[PATCH v4 0/9] fs: interface for directly reading/writing compressed data

[PATCH v4 0/9] fs: interface for directly reading/writing compressed data

[Omar Sandoval]

From: Omar Sandoval <osandov@fb.com>

Hello,

This series adds an API for reading compressed data on a filesystem
without decompressing it as well as support for writing compressed data
directly to the filesystem. As with the previous submissions, I've
included a man page patch describing the API. I have test cases
(including fsstress support) and example programs which I'll send up
once the interface is more or less settled [1].

The main use-case is Btrfs send/receive: currently, when sending data
from one compressed filesystem to another, the sending side decompresses
the data and the receiving side recompresses it before writing it out.
This is wasteful and can be avoided if we can just send and write
compressed extents. The send part will be implemented in a separate
series, as this API can stand alone.

I'm fairly happy with the UAPI and VFS interface now. I'd love for Al
and/or Christoph to take a look at that part. The Btrfs side is mostly
there, just missing read repair.

Patches 1-3 add the VFS support. Patches 4-7 are Btrfs prep patches.
Patch 8 adds Btrfs encoded read support and patch 9 adds Btrfs encoded
write support.

Changes from v3:

- Rebase on v5.6-rc3.
- Disallow extents with a file length greater than the unencoded
  length.
- Drop Btrfs cleanups and fixes that have already been merged.
- Add Nikolay's reviewed-bys.

Please share any comments on the API or implementation. Thanks!

1: https://github.com/osandov/xfstests/tree/rwf-encoded
2: https://lore.kernel.org/linux-btrfs/cover.1574273658.git.osandov@fb.com/

Omar Sandoval (9):
  iov_iter: add copy_struct_from_iter()
  fs: add O_ALLOW_ENCODED open flag
  fs: add RWF_ENCODED for reading/writing compressed data
  btrfs: don't advance offset for compressed bios in btrfs_csum_one_bio()
  btrfs: add ram_bytes and offset to btrfs_ordered_extent
  btrfs: support different disk extent size for delalloc
  btrfs: optionally extend i_size in cow_file_range_inline()
  btrfs: implement RWF_ENCODED reads
  btrfs: implement RWF_ENCODED writes

 Documentation/filesystems/encoded_io.rst |  74 ++
 Documentation/filesystems/index.rst      |   1 +
 arch/alpha/include/uapi/asm/fcntl.h      |   1 +
 arch/parisc/include/uapi/asm/fcntl.h     |   1 +
 arch/sparc/include/uapi/asm/fcntl.h      |   1 +
 fs/btrfs/compression.c                   |  12 +-
 fs/btrfs/compression.h                   |   6 +-
 fs/btrfs/ctree.h                         |   9 +-
 fs/btrfs/delalloc-space.c                |  38 +-
 fs/btrfs/delalloc-space.h                |   4 +-
 fs/btrfs/file-item.c                     |  35 +-
 fs/btrfs/file.c                          |  55 +-
 fs/btrfs/inode.c                         | 873 ++++++++++++++++++++---
 fs/btrfs/ordered-data.c                  |  77 +-
 fs/btrfs/ordered-data.h                  |  18 +-
 fs/btrfs/relocation.c                    |   4 +-
 fs/fcntl.c                               |  10 +-
 fs/namei.c                               |   4 +
 include/linux/fcntl.h                    |   2 +-
 include/linux/fs.h                       |  16 +
 include/linux/uio.h                      |   2 +
 include/uapi/asm-generic/fcntl.h         |   4 +
 include/uapi/linux/fs.h                  |  33 +-
 lib/iov_iter.c                           |  82 +++
 mm/filemap.c                             | 166 ++++-
 25 files changed, 1306 insertions(+), 222 deletions(-)
 create mode 100644 Documentation/filesystems/encoded_io.rst

-- 
2.25.1

[PATCH man-pages v4] Document encoded I/O

[Omar Sandoval]

From: Omar Sandoval <osandov@fb.com>

This adds a new page, encoded_io(7), providing an overview of encoded
I/O and updates fcntl(2), open(2), and preadv2(2)/pwritev2(2) to
reference it.

Signed-off-by: Omar Sandoval <osandov@fb.com>
---
 man2/fcntl.2      |  10 +-
 man2/open.2       |  13 ++
 man2/readv.2      |  64 ++++++++++
 man7/encoded_io.7 | 298 ++++++++++++++++++++++++++++++++++++++++++++++
 4 files changed, 384 insertions(+), 1 deletion(-)
 create mode 100644 man7/encoded_io.7

diff --git a/man2/fcntl.2 b/man2/fcntl.2
index bb1ac1f5d..15a1010a6 100644
--- a/man2/fcntl.2
+++ b/man2/fcntl.2
@@ -222,8 +222,9 @@ On Linux, this command can change only the
 .BR O_ASYNC ,
 .BR O_DIRECT ,
 .BR O_NOATIME ,
+.BR O_NONBLOCK ,
 and
-.B O_NONBLOCK
+.B O_ALLOW_ENCODED
 flags.
 It is not possible to change the
 .BR O_DSYNC
@@ -1821,6 +1822,13 @@ Attempted to clear the
 flag on a file that has the append-only attribute set.
 .TP
 .B EPERM
+Attempted to set the
+.B O_ALLOW_ENCODED
+flag and the calling process did not have the
+.B CAP_SYS_ADMIN
+capability.
+.TP
+.B EPERM
 .I cmd
 was
 .BR F_ADD_SEALS ,
diff --git a/man2/open.2 b/man2/open.2
index 3ab4ee17b..256cb4247 100644
--- a/man2/open.2
+++ b/man2/open.2
@@ -421,6 +421,14 @@ was followed by a call to
 .BR fdatasync (2)).
 .IR "See NOTES below" .
 .TP
+.B O_ALLOW_ENCODED
+Open the file with encoded I/O permissions;
+see
+.BR encoded_io (7).
+The caller must have the
+.B CAP_SYS_ADMIN
+capability.
+.TP
 .B O_EXCL
 Ensure that this call creates the file:
 if this flag is specified in conjunction with
@@ -1176,6 +1184,11 @@ did not match the owner of the file and the caller was not privileged.
 The operation was prevented by a file seal; see
 .BR fcntl (2).
 .TP
+.B EPERM
+The
+.B O_ALLOW_ENCODED
+flag was specified, but the caller was not privileged.
+.TP
 .B EROFS
 .I pathname
 refers to a file on a read-only filesystem and write access was
diff --git a/man2/readv.2 b/man2/readv.2
index af27aa63e..8b5458023 100644
--- a/man2/readv.2
+++ b/man2/readv.2
@@ -265,6 +265,11 @@ the data is always appended to the end of the file.
 However, if the
 .I offset
 argument is \-1, the current file offset is updated.
+.TP
+.BR RWF_ENCODED " (since Linux 5.7)"
+Read or write encoded (e.g., compressed) data.
+See
+.BR encoded_io (7).
 .SH RETURN VALUE
 On success,
 .BR readv (),
@@ -284,6 +289,13 @@ than requested (see
 and
 .BR write (2)).
 .PP
+If
+.B
+RWF_ENCODED
+was specified in
+.IR flags ,
+then the return value is the number of encoded bytes.
+.PP
 On error, \-1 is returned, and \fIerrno\fP is set appropriately.
 .SH ERRORS
 The errors are as given for
@@ -314,6 +326,58 @@ is less than zero or greater than the permitted maximum.
 .TP
 .B EOPNOTSUPP
 An unknown flag is specified in \fIflags\fP.
+.TP
+.B EOPNOTSUPP
+.B RWF_ENCODED
+is specified in
+.I flags
+and the filesystem does not implement encoded I/O.
+.TP
+.B EPERM
+.B RWF_ENCODED
+is specified in
+.I flags
+and the file was not opened with the
+.B O_ALLOW_ENCODED
+flag.
+.PP
+.BR preadv2 ()
+can fail for the following reasons:
+.TP
+.B E2BIG
+.B RWF_ENCODED
+is specified in
+.I flags
+and
+.I iov[0]
+is not large enough to return the encoding metadata.
+.TP
+.B ENOBUFS
+.B RWF_ENCODED
+is specified in
+.I flags
+and the buffers in
+.I iov
+are not big enough to return the encoded data.
+.PP
+.BR pwritev2 ()
+can fail for the following reasons:
+.TP
+.B E2BIG
+.B RWF_ENCODED
+is specified in
+.I flags
+and
+.I iov[0]
+contains non-zero fields
+after the kernel's
+.IR "sizeof(struct\ encoded_iov)" .
+.TP
+.B EINVAL
+.B RWF_ENCODED
+is specified in
+.I flags
+and the alignment and/or size requirements are not met.
 .SH VERSIONS
 .BR preadv ()
 and
diff --git a/man7/encoded_io.7 b/man7/encoded_io.7
new file mode 100644
index 000000000..72b40353f
--- /dev/null
+++ b/man7/encoded_io.7
@@ -0,0 +1,298 @@
+.\" Copyright (c) 2019 by Omar Sandoval <osandov@fb.com>
+.\"
+.\" %%%LICENSE_START(VERBATIM)
+.\" Permission is granted to make and distribute verbatim copies of this
+.\" manual provided the copyright notice and this permission notice are
+.\" preserved on all copies.
+.\"
+.\" Permission is granted to copy and distribute modified versions of this
+.\" manual under the conditions for verbatim copying, provided that the
+.\" entire resulting derived work is distributed under the terms of a
+.\" permission notice identical to this one.
+.\"
+.\" Since the Linux kernel and libraries are constantly changing, this
+.\" manual page may be incorrect or out-of-date.  The author(s) assume no
+.\" responsibility for errors or omissions, or for damages resulting from
+.\" the use of the information contained herein.  The author(s) may not
+.\" have taken the same level of care in the production of this manual,
+.\" which is licensed free of charge, as they might when working
+.\" professionally.
+.\"
+.\" Formatted or processed versions of this manual, if unaccompanied by
+.\" the source, must acknowledge the copyright and authors of this work.
+.\" %%%LICENSE_END
+.\"
+.\"
+.TH ENCODED_IO  7 2019-10-14 "Linux" "Linux Programmer's Manual"
+.SH NAME
+encoded_io \- overview of encoded I/O
+.SH DESCRIPTION
+Several filesystems (e.g., Btrfs) support transparent encoding
+(e.g., compression, encryption) of data on disk:
+written data is encoded by the kernel before it is written to disk,
+and read data is decoded before being returned to the user.
+In some cases, it is useful to skip this encoding step.
+For example, the user may want to read the compressed contents of a file
+or write pre-compressed data directly to a file.
+This is referred to as "encoded I/O".
+.SS Encoded I/O API
+Encoded I/O is specified with the
+.B RWF_ENCODED
+flag to
+.BR preadv2 (2)
+and
+.BR pwritev2 (2).
+If
+.B RWF_ENCODED
+is specified, then
+.I iov[0].iov_base
+points to an
+.I
+encoded_iov
+structure, defined in
+.I <linux/fs.h>
+as:
+.PP
+.in +4n
+.EX
+struct encoded_iov {
+    __aligned_u64 len;
+    __aligned_u64 unencoded_len;
+    __aligned_u64 unencoded_offset;
+    __u32 compression;
+    __u32 encryption;
+};
+.EE
+.in
+.PP
+This may be extended in the future, so
+.I iov[0].iov_len
+must be set to
+.I "sizeof(struct\ encoded_iov)"
+for forward/backward compatibility.
+The remaining buffers contain the encoded data.
+.PP
+.I compression
+and
+.I encryption
+are the encoding fields.
+.I compression
+is one of
+.B ENCODED_IOV_COMPRESSION_NONE
+(zero),
+.BR ENCODED_IOV_COMPRESSION_ZLIB ,
+.BR ENCODED_IOV_COMPRESSION_LZO ,
+or
+.BR ENCODED_IOV_COMPRESSION_ZSTD .
+.I encryption
+is currently always
+.B ENCODED_IOV_ENCRYPTION_NONE
+(zero).
+.PP
+.I unencoded_len
+is the length of the unencoded (i.e., decrypted and decompressed) data.
+.I unencoded_offset
+is the offset into the unencoded data where the data in the file begins
+(less than or equal to
+.IR unencoded_len ).
+.I len
+is the length of the data in the file
+(less than or equal to
+.I unencoded_len
+-
+.IR unencoded_offset ).
+.I
+.PP
+In most cases,
+.I len
+is equal to
+.I unencoded_len
+and
+.I unencoded_offset
+is zero.
+However, it may be necessary to refer to a subset of the unencoded data,
+usually because a read occurred in the middle of an encoded extent,
+because part of an extent was overwritten or deallocated in some
+way (e.g., with
+.BR write (2),
+.BR truncate (2),
+or
+.BR fallocate (2))
+or because part of an extent was added to the file (e.g., with
+.BR ioctl_ficlonerange (2)
+or
+.BR ioctl_fideduperange (2)).
+For example, if
+.I len
+is 300,
+.I unencoded_len
+is 1000,
+and
+.I unencoded_offset
+is 600,
+then the encoded data is 1000 bytes long when decoded,
+of which only the 300 bytes starting at offset 600 are used;
+the first 600 and last 100 bytes should be ignored.
+.PP
+If the unencoded data is actually longer than
+.IR unencoded_len ,
+then it is truncated;
+if it is shorter, then it is extended with zeroes.
+.PP
+For
+.BR pwritev2 (),
+the metadata should be specified in
+.IR iov[0] .
+If
+.I iov[0].iov_len
+is less than
+.I "sizeof(struct\ encoded_iov)"
+in the kernel,
+then any fields unknown to userspace are treated as if they were zero;
+if it is greater and any fields unknown to the kernel are non-zero,
+then this returns -1 and sets
+.I errno
+to
+.BR E2BIG .
+The encoded data should be passed in the remaining buffers.
+This returns the number of encoded bytes written (that is, the sum of
+.I iov[n].iov_len
+for 1 <=
+.I n
+<
+.IR iovcnt ;
+partial writes will not occur).
+If the
+.I offset
+argument to
+.BR pwritev2 ()
+is -1, then the file offset is incremented by
+.IR len .
+At least one encoding field must be non-zero.
+Note that the encoded data is not validated when it is written;
+if it is not valid (e.g., it cannot be decompressed),
+then a subsequent read may return an error.
+.PP
+For
+.BR preadv2 (),
+the metadata is returned in
+.IR iov[0] .
+If
+.I iov[0].iov_len
+is less than
+.I "sizeof(struct\ encoded_iov)"
+in the kernel and any fields unknown to userspace are non-zero,
+then this returns -1 and sets
+.I errno
+to
+.BR E2BIG ;
+if it is greater,
+then any fields unknown to the kernel are returned as zero.
+The encoded data is returned in the remaining buffers.
+If the provided buffers are not large enough to return an entire encoded
+extent,
+then this returns -1 and sets
+.I errno
+to
+.BR ENOBUFS .
+This returns the number of encoded bytes read.
+If the
+.I offset
+argument to
+.BR preadv2 ()
+is -1, then the file offset is incremented by
+.IR len .
+This will only return one encoded extent per call.
+This can also read data which is not encoded;
+all encoding fields will be zero in that case.
+.PP
+As the filesystem page cache typically contains decoded data,
+encoded I/O bypasses the page cache.
+.SS Security
+Encoded I/O creates the potential for some security issues:
+.IP * 3
+Encoded writes allow writing arbitrary data which the kernel will decode on
+a subsequent read. Decompression algorithms are complex and may have bugs
+which can be exploited by maliciously crafted data.
+.IP *
+Encoded reads may return data which is not logically present in the file
+(see the discussion of
+.I len
+vs.
+.I unencoded_len
+above).
+It may not be intended for this data to be readable.
+.PP
+Therefore, encoded I/O requires privilege.
+Namely, the
+.B RWF_ENCODED
+flag may only be used when the file was opened with the
+.B O_ALLOW_ENCODED
+flag to
+.BR open (2),
+which requires the
+.B CAP_SYS_ADMIN
+capability.
+.B O_ALLOW_ENCODED
+may be set and cleared with
+.BR fcntl (2).
+Note that it is not cleared on
+.BR fork (2)
+or
+.BR execve (2);
+one may wish to use
+.B O_CLOEXEC
+with
+.BR O_ALLOW_ENCODED .
+.SS Filesystem support
+Encoded I/O is supported on the following filesystems:
+.TP
+Btrfs (since Linux 5.8)
+.IP
+Btrfs supports encoded reads and writes of compressed data.
+The data is encoded as follows:
+.RS
+.IP * 3
+If
+.I compression
+is
+.BR ENCODED_IOV_COMPRESSION_ZLIB ,
+then the encoded data is a single zlib stream.
+.IP *
+If
+.I compression
+is
+.BR ENCODED_IOV_COMPRESSION_LZO ,
+then the encoded data is compressed page by page with LZO1X
+and wrapped in the format documented in the Linux kernel source file
+.IR fs/btrfs/lzo.c .
+.IP *
+If
+.I compression
+is
+.BR ENCODED_IOV_COMPRESSION_ZSTD ,
+then the encoded data is a single zstd frame compressed with the
+.I windowLog
+compression parameter set to no more than 17.
+.RE
+.IP
+Additionally, there are some restrictions on
+.BR pwritev2 ():
+.RS
+.IP * 3
+.I offset
+(or the current file offset if
+.I offset
+is -1) must be aligned to the sector size of the filesystem.
+.IP *
+.I len
+must be aligned to the sector size of the filesystem
+unless the data ends at or beyond the current end of the file.
+.IP *
+.I unencoded_len
+and the length of the encoded data must each be no more than 128 KiB.
+This limit may increase in the future.
+.IP *
+The length of the encoded data must be less than or equal to
+.IR unencoded_len .
+.RE
-- 
2.25.1

Re: [PATCH man-pages v4] Document encoded I/O

[Amir Goldstein]

> +encoded_io \- overview of encoded I/O
> +.SH DESCRIPTION
> +Several filesystems (e.g., Btrfs) support transparent encoding
> +(e.g., compression, encryption) of data on disk:
> +written data is encoded by the kernel before it is written to disk,
> +and read data is decoded before being returned to the user.
> +In some cases, it is useful to skip this encoding step.
> +For example, the user may want to read the compressed contents of a file
> +or write pre-compressed data directly to a file.
> +This is referred to as "encoded I/O".
> +.SS Encoded I/O API
> +Encoded I/O is specified with the
> +.B RWF_ENCODED
> +flag to
> +.BR preadv2 (2)
> +and
> +.BR pwritev2 (2).
> +If
> +.B RWF_ENCODED
> +is specified, then
> +.I iov[0].iov_base
> +points to an
> +.I
> +encoded_iov
> +structure, defined in
> +.I <linux/fs.h>
> +as:
> +.PP
> +.in +4n
> +.EX
> +struct encoded_iov {
> +    __aligned_u64 len;
> +    __aligned_u64 unencoded_len;
> +    __aligned_u64 unencoded_offset;
> +    __u32 compression;
> +    __u32 encryption;
> +};

This new API can generate many diverse error conditions that the standard errno
codes are not rich enough to describe.
Maybe add room for encoded io specific error codes in the metadata structure
would be good practice, for example:
- compression method not supported
- encryption method not supported
- the combination of enc/comp is not supported
- and so on


> +.EE
> +.in
> +.PP
> +This may be extended in the future, so
> +.I iov[0].iov_len
> +must be set to
> +.I "sizeof(struct\ encoded_iov)"
> +for forward/backward compatibility.
> +The remaining buffers contain the encoded data.
> +.PP
> +.I compression
> +and
> +.I encryption
> +are the encoding fields.
> +.I compression
> +is one of
> +.B ENCODED_IOV_COMPRESSION_NONE
> +(zero),
> +.BR ENCODED_IOV_COMPRESSION_ZLIB ,
> +.BR ENCODED_IOV_COMPRESSION_LZO ,
> +or
> +.BR ENCODED_IOV_COMPRESSION_ZSTD .
> +.I encryption
> +is currently always
> +.B ENCODED_IOV_ENCRYPTION_NONE
> +(zero).
> +.PP
> +.I unencoded_len
> +is the length of the unencoded (i.e., decrypted and decompressed) data.
> +.I unencoded_offset
> +is the offset into the unencoded data where the data in the file begins
> +(less than or equal to
> +.IR unencoded_len ).
> +.I len
> +is the length of the data in the file
> +(less than or equal to
> +.I unencoded_len
> +-
> +.IR unencoded_offset ).
> +.I
> +.PP
> +In most cases,
> +.I len
> +is equal to
> +.I unencoded_len
> +and
> +.I unencoded_offset
> +is zero.
> +However, it may be necessary to refer to a subset of the unencoded data,
> +usually because a read occurred in the middle of an encoded extent,
> +because part of an extent was overwritten or deallocated in some
> +way (e.g., with
> +.BR write (2),
> +.BR truncate (2),
> +or
> +.BR fallocate (2))
> +or because part of an extent was added to the file (e.g., with
> +.BR ioctl_ficlonerange (2)
> +or
> +.BR ioctl_fideduperange (2)).
> +For example, if
> +.I len
> +is 300,
> +.I unencoded_len
> +is 1000,
> +and
> +.I unencoded_offset
> +is 600,
> +then the encoded data is 1000 bytes long when decoded,
> +of which only the 300 bytes starting at offset 600 are used;
> +the first 600 and last 100 bytes should be ignored.
> +.PP
> +If the unencoded data is actually longer than
> +.IR unencoded_len ,
> +then it is truncated;
> +if it is shorter, then it is extended with zeroes.

I find the unencoded_len/unencoded_offset API extremely confusing and all
the clarifications above did not help to ease this feeling.
Please remind me why does the API need to expose unencoded details at all.
I understand the backup/restore use case for read/write encoded data.
I do not understand how unencoded offset info is relevant to this use case
or what are the other use cases it is relevant for.

> +.PP
> +For
> +.BR pwritev2 (),
> +the metadata should be specified in
> +.IR iov[0] .
> +If
> +.I iov[0].iov_len
> +is less than
> +.I "sizeof(struct\ encoded_iov)"
> +in the kernel,
> +then any fields unknown to userspace are treated as if they were zero;
> +if it is greater and any fields unknown to the kernel are non-zero,
> +then this returns -1 and sets
> +.I errno
> +to
> +.BR E2BIG .
> +The encoded data should be passed in the remaining buffers.
> +This returns the number of encoded bytes written (that is, the sum of
> +.I iov[n].iov_len
> +for 1 <=
> +.I n
> +<
> +.IR iovcnt ;
> +partial writes will not occur).
> +If the
> +.I offset
> +argument to
> +.BR pwritev2 ()
> +is -1, then the file offset is incremented by
> +.IR len .
> +At least one encoding field must be non-zero.
> +Note that the encoded data is not validated when it is written;
> +if it is not valid (e.g., it cannot be decompressed),
> +then a subsequent read may return an error.
> +.PP
> +For
> +.BR preadv2 (),
> +the metadata is returned in
> +.IR iov[0] .
> +If
> +.I iov[0].iov_len
> +is less than
> +.I "sizeof(struct\ encoded_iov)"
> +in the kernel and any fields unknown to userspace are non-zero,
> +then this returns -1 and sets
> +.I errno
> +to
> +.BR E2BIG ;
> +if it is greater,
> +then any fields unknown to the kernel are returned as zero.
> +The encoded data is returned in the remaining buffers.
> +If the provided buffers are not large enough to return an entire encoded
> +extent,
> +then this returns -1 and sets
> +.I errno
> +to
> +.BR ENOBUFS .
> +This returns the number of encoded bytes read.
> +If the
> +.I offset
> +argument to
> +.BR preadv2 ()
> +is -1, then the file offset is incremented by
> +.IR len .
> +This will only return one encoded extent per call.
> +This can also read data which is not encoded;
> +all encoding fields will be zero in that case.
> +.PP
> +As the filesystem page cache typically contains decoded data,
> +encoded I/O bypasses the page cache.
> +.SS Security
> +Encoded I/O creates the potential for some security issues:
> +.IP * 3
> +Encoded writes allow writing arbitrary data which the kernel will decode on
> +a subsequent read. Decompression algorithms are complex and may have bugs
> +which can be exploited by maliciously crafted data.
> +.IP *
> +Encoded reads may return data which is not logically present in the file
> +(see the discussion of
> +.I len
> +vs.
> +.I unencoded_len
> +above).
> +It may not be intended for this data to be readable.
> +.PP
> +Therefore, encoded I/O requires privilege.
> +Namely, the
> +.B RWF_ENCODED
> +flag may only be used when the file was opened with the
> +.B O_ALLOW_ENCODED
> +flag to
> +.BR open (2),
> +which requires the
> +.B CAP_SYS_ADMIN
> +capability.
> +.B O_ALLOW_ENCODED
> +may be set and cleared with
> +.BR fcntl (2).
> +Note that it is not cleared on
> +.BR fork (2)
> +or
> +.BR execve (2);
> +one may wish to use
> +.B O_CLOEXEC
> +with
> +.BR O_ALLOW_ENCODED .

Sigh! If I were an attacker I would be drooling right now.
We want to create a new API to read/write raw encrypted data (even though
you have not implemented any encryption yet) and we use the same old
vulnerable practices that security people have been fighting for decades?
I am not very comfortable with this attitude.
I think we should be much more prudent for the first version of the API.

How about not allowing to set O_ALLOW_ENCODED without O_CLOEXEC.
We may or may not allow to clear O_CLOEXEC while O_ALLOW_ENCODED
is set, in case this is the user intention, but leaving the API as it is is just
asking for trouble IMO.

Thanks,
Amir.

Re: [PATCH man-pages v4] Document encoded I/O

[Omar Sandoval]

On Sat, Feb 29, 2020 at 12:28:41PM +0200, Amir Goldstein wrote:
> > +encoded_io \- overview of encoded I/O
> > +.SH DESCRIPTION
> > +Several filesystems (e.g., Btrfs) support transparent encoding
> > +(e.g., compression, encryption) of data on disk:
> > +written data is encoded by the kernel before it is written to disk,
> > +and read data is decoded before being returned to the user.
> > +In some cases, it is useful to skip this encoding step.
> > +For example, the user may want to read the compressed contents of a file
> > +or write pre-compressed data directly to a file.
> > +This is referred to as "encoded I/O".
> > +.SS Encoded I/O API
> > +Encoded I/O is specified with the
> > +.B RWF_ENCODED
> > +flag to
> > +.BR preadv2 (2)
> > +and
> > +.BR pwritev2 (2).
> > +If
> > +.B RWF_ENCODED
> > +is specified, then
> > +.I iov[0].iov_base
> > +points to an
> > +.I
> > +encoded_iov
> > +structure, defined in
> > +.I <linux/fs.h>
> > +as:
> > +.PP
> > +.in +4n
> > +.EX
> > +struct encoded_iov {
> > +    __aligned_u64 len;
> > +    __aligned_u64 unencoded_len;
> > +    __aligned_u64 unencoded_offset;
> > +    __u32 compression;
> > +    __u32 encryption;
> > +};
> 
> This new API can generate many diverse error conditions that the standard errno
> codes are not rich enough to describe.
> Maybe add room for encoded io specific error codes in the metadata structure
> would be good practice, for example:
> - compression method not supported
> - encryption method not supported
> - the combination of enc/comp is not supported
> - and so on

I like this idea, but it feels like even more iovec abuse. Namely, for
pwritev2(), it feels a little off that we'd be copying _to_ user memory
rather than only copying from. It's probably worth it for better errors,
though.

> > +.EE
> > +.in
> > +.PP
> > +This may be extended in the future, so
> > +.I iov[0].iov_len
> > +must be set to
> > +.I "sizeof(struct\ encoded_iov)"
> > +for forward/backward compatibility.
> > +The remaining buffers contain the encoded data.
> > +.PP
> > +.I compression
> > +and
> > +.I encryption
> > +are the encoding fields.
> > +.I compression
> > +is one of
> > +.B ENCODED_IOV_COMPRESSION_NONE
> > +(zero),
> > +.BR ENCODED_IOV_COMPRESSION_ZLIB ,
> > +.BR ENCODED_IOV_COMPRESSION_LZO ,
> > +or
> > +.BR ENCODED_IOV_COMPRESSION_ZSTD .
> > +.I encryption
> > +is currently always
> > +.B ENCODED_IOV_ENCRYPTION_NONE
> > +(zero).
> > +.PP
> > +.I unencoded_len
> > +is the length of the unencoded (i.e., decrypted and decompressed) data.
> > +.I unencoded_offset
> > +is the offset into the unencoded data where the data in the file begins
> > +(less than or equal to
> > +.IR unencoded_len ).
> > +.I len
> > +is the length of the data in the file
> > +(less than or equal to
> > +.I unencoded_len
> > +-
> > +.IR unencoded_offset ).
> > +.I
> > +.PP
> > +In most cases,
> > +.I len
> > +is equal to
> > +.I unencoded_len
> > +and
> > +.I unencoded_offset
> > +is zero.
> > +However, it may be necessary to refer to a subset of the unencoded data,
> > +usually because a read occurred in the middle of an encoded extent,
> > +because part of an extent was overwritten or deallocated in some
> > +way (e.g., with
> > +.BR write (2),
> > +.BR truncate (2),
> > +or
> > +.BR fallocate (2))
> > +or because part of an extent was added to the file (e.g., with
> > +.BR ioctl_ficlonerange (2)
> > +or
> > +.BR ioctl_fideduperange (2)).
> > +For example, if
> > +.I len
> > +is 300,
> > +.I unencoded_len
> > +is 1000,
> > +and
> > +.I unencoded_offset
> > +is 600,
> > +then the encoded data is 1000 bytes long when decoded,
> > +of which only the 300 bytes starting at offset 600 are used;
> > +the first 600 and last 100 bytes should be ignored.
> > +.PP
> > +If the unencoded data is actually longer than
> > +.IR unencoded_len ,
> > +then it is truncated;
> > +if it is shorter, then it is extended with zeroes.
> 
> I find the unencoded_len/unencoded_offset API extremely confusing and all
> the clarifications above did not help to ease this feeling.
> Please remind me why does the API need to expose unencoded details at all.
> I understand the backup/restore use case for read/write encoded data.
> I do not understand how unencoded offset info is relevant to this use case
> or what are the other use cases it is relevant for.

I agree, it's confusing. However, without this concept on the read side,
there's no way to represent some file extent layouts, and without the
write side, those layouts can't be written back out. That would make
this interface much less useful for backups.

These cases arise in a few ways on Btrfs:

1. Files with a size unaligned to the block size.

   Ignoring inline data, Btrfs always pads data to the filesystem block
   size when compressing. So, a file with a size unaligned to the block
   size will end with an extent that decompresses to a multiple of the
   block size, but logically the file only contains the data up to
   i_size. In this case, len (length up to i_size) < unencoded_len (full
   decompressed length). This can arise simply from writing out an
   unaligned file or from truncating a file unaligned.

2. FICLONERANGE from the middle of an extent.

   Suppose file A has a large compressed extent with
   len = unencoded_len = 128k and unencoded_offset = 0. If the user does
   an FICLONERANGE out of the middle of that extent (say, 64k long and
   4k from the start of the extent), Btrfs creates a "partial" extent
   which references the original extent (in my example, the result would
   have len = 64k, unencoded_offset = 4k, and unencoded_len still 128k).

3. Overwriting the middle of an extent.

   In some cases, when the middle of an extent is overwritten (e.g., an
   O_DIRECT write, FICLONERANGE, or FIDEDUPERANGE), Btrfs splits up the
   overwritten extents into partial extents referencing the original
   extent instead of rewriting the whole extent.

These aren't specific to compression or Btrfs' on-disk format. fscrypt
uses block ciphers for file data, so case 1 is just as relevant for
that. The way Btrfs handles case 2 is the only sane way I can see for
supporting FICLONERANGE for encoded data.

> > +.PP
> > +For
> > +.BR pwritev2 (),
> > +the metadata should be specified in
> > +.IR iov[0] .
> > +If
> > +.I iov[0].iov_len
> > +is less than
> > +.I "sizeof(struct\ encoded_iov)"
> > +in the kernel,
> > +then any fields unknown to userspace are treated as if they were zero;
> > +if it is greater and any fields unknown to the kernel are non-zero,
> > +then this returns -1 and sets
> > +.I errno
> > +to
> > +.BR E2BIG .
> > +The encoded data should be passed in the remaining buffers.
> > +This returns the number of encoded bytes written (that is, the sum of
> > +.I iov[n].iov_len
> > +for 1 <=
> > +.I n
> > +<
> > +.IR iovcnt ;
> > +partial writes will not occur).
> > +If the
> > +.I offset
> > +argument to
> > +.BR pwritev2 ()
> > +is -1, then the file offset is incremented by
> > +.IR len .
> > +At least one encoding field must be non-zero.
> > +Note that the encoded data is not validated when it is written;
> > +if it is not valid (e.g., it cannot be decompressed),
> > +then a subsequent read may return an error.
> > +.PP
> > +For
> > +.BR preadv2 (),
> > +the metadata is returned in
> > +.IR iov[0] .
> > +If
> > +.I iov[0].iov_len
> > +is less than
> > +.I "sizeof(struct\ encoded_iov)"
> > +in the kernel and any fields unknown to userspace are non-zero,
> > +then this returns -1 and sets
> > +.I errno
> > +to
> > +.BR E2BIG ;
> > +if it is greater,
> > +then any fields unknown to the kernel are returned as zero.
> > +The encoded data is returned in the remaining buffers.
> > +If the provided buffers are not large enough to return an entire encoded
> > +extent,
> > +then this returns -1 and sets
> > +.I errno
> > +to
> > +.BR ENOBUFS .
> > +This returns the number of encoded bytes read.
> > +If the
> > +.I offset
> > +argument to
> > +.BR preadv2 ()
> > +is -1, then the file offset is incremented by
> > +.IR len .
> > +This will only return one encoded extent per call.
> > +This can also read data which is not encoded;
> > +all encoding fields will be zero in that case.
> > +.PP
> > +As the filesystem page cache typically contains decoded data,
> > +encoded I/O bypasses the page cache.
> > +.SS Security
> > +Encoded I/O creates the potential for some security issues:
> > +.IP * 3
> > +Encoded writes allow writing arbitrary data which the kernel will decode on
> > +a subsequent read. Decompression algorithms are complex and may have bugs
> > +which can be exploited by maliciously crafted data.
> > +.IP *
> > +Encoded reads may return data which is not logically present in the file
> > +(see the discussion of
> > +.I len
> > +vs.
> > +.I unencoded_len
> > +above).
> > +It may not be intended for this data to be readable.
> > +.PP
> > +Therefore, encoded I/O requires privilege.
> > +Namely, the
> > +.B RWF_ENCODED
> > +flag may only be used when the file was opened with the
> > +.B O_ALLOW_ENCODED
> > +flag to
> > +.BR open (2),
> > +which requires the
> > +.B CAP_SYS_ADMIN
> > +capability.
> > +.B O_ALLOW_ENCODED
> > +may be set and cleared with
> > +.BR fcntl (2).
> > +Note that it is not cleared on
> > +.BR fork (2)
> > +or
> > +.BR execve (2);
> > +one may wish to use
> > +.B O_CLOEXEC
> > +with
> > +.BR O_ALLOW_ENCODED .
> 
> Sigh! If I were an attacker I would be drooling right now.
> We want to create a new API to read/write raw encrypted data (even though
> you have not implemented any encryption yet) and we use the same old
> vulnerable practices that security people have been fighting for decades?
> I am not very comfortable with this attitude.
> I think we should be much more prudent for the first version of the API.
> 
> How about not allowing to set O_ALLOW_ENCODED without O_CLOEXEC.
> We may or may not allow to clear O_CLOEXEC while O_ALLOW_ENCODED
> is set, in case this is the user intention, but leaving the API as it is is just
> asking for trouble IMO.

Ok, I'm fine with requiring O_CLOEXEC for O_ALLOW_ENCODED on open. I'm
pretty sure we want to allow clearing it with fcntl, as that is a very
intentional action.

Re: [PATCH man-pages v4] Document encoded I/O

[Amir Goldstein]

On Sat, Feb 29, 2020 at 8:03 PM Omar Sandoval <osandov@osandov.com> wrote:
>
> On Sat, Feb 29, 2020 at 12:28:41PM +0200, Amir Goldstein wrote:
> > > +encoded_io \- overview of encoded I/O
> > > +.SH DESCRIPTION
> > > +Several filesystems (e.g., Btrfs) support transparent encoding
> > > +(e.g., compression, encryption) of data on disk:
> > > +written data is encoded by the kernel before it is written to disk,
> > > +and read data is decoded before being returned to the user.
> > > +In some cases, it is useful to skip this encoding step.
> > > +For example, the user may want to read the compressed contents of a file
> > > +or write pre-compressed data directly to a file.
> > > +This is referred to as "encoded I/O".
> > > +.SS Encoded I/O API
> > > +Encoded I/O is specified with the
> > > +.B RWF_ENCODED
> > > +flag to
> > > +.BR preadv2 (2)
> > > +and
> > > +.BR pwritev2 (2).
> > > +If
> > > +.B RWF_ENCODED
> > > +is specified, then
> > > +.I iov[0].iov_base
> > > +points to an
> > > +.I
> > > +encoded_iov
> > > +structure, defined in
> > > +.I <linux/fs.h>
> > > +as:
> > > +.PP
> > > +.in +4n
> > > +.EX
> > > +struct encoded_iov {
> > > +    __aligned_u64 len;
> > > +    __aligned_u64 unencoded_len;
> > > +    __aligned_u64 unencoded_offset;
> > > +    __u32 compression;
> > > +    __u32 encryption;
> > > +};
> >
> > This new API can generate many diverse error conditions that the standard errno
> > codes are not rich enough to describe.
> > Maybe add room for encoded io specific error codes in the metadata structure
> > would be good practice, for example:
> > - compression method not supported
> > - encryption method not supported
> > - the combination of enc/comp is not supported
> > - and so on
>
> I like this idea, but it feels like even more iovec abuse. Namely, for

That's true.

> pwritev2(), it feels a little off that we'd be copying _to_ user memory
> rather than only copying from. It's probably worth it for better errors,
> though.
>

Apropos iovec abuse, if encoded io is going to interpret iovec[0] differently
why not interpret iovec arg differently. The result might be less awkward if
the structure passed to preadv2/pwritev2 is struct encoded_iov * instead
of struct iov *.

> > > +.EE
> > > +.in
> > > +.PP
> > > +This may be extended in the future, so
> > > +.I iov[0].iov_len
> > > +must be set to
> > > +.I "sizeof(struct\ encoded_iov)"
> > > +for forward/backward compatibility.
> > > +The remaining buffers contain the encoded data.
> > > +.PP
> > > +.I compression
> > > +and
> > > +.I encryption
> > > +are the encoding fields.
> > > +.I compression
> > > +is one of
> > > +.B ENCODED_IOV_COMPRESSION_NONE
> > > +(zero),
> > > +.BR ENCODED_IOV_COMPRESSION_ZLIB ,
> > > +.BR ENCODED_IOV_COMPRESSION_LZO ,
> > > +or
> > > +.BR ENCODED_IOV_COMPRESSION_ZSTD .
> > > +.I encryption
> > > +is currently always
> > > +.B ENCODED_IOV_ENCRYPTION_NONE
> > > +(zero).
> > > +.PP
> > > +.I unencoded_len
> > > +is the length of the unencoded (i.e., decrypted and decompressed) data.
> > > +.I unencoded_offset
> > > +is the offset into the unencoded data where the data in the file begins
> > > +(less than or equal to
> > > +.IR unencoded_len ).
> > > +.I len
> > > +is the length of the data in the file
> > > +(less than or equal to
> > > +.I unencoded_len
> > > +-
> > > +.IR unencoded_offset ).
> > > +.I
> > > +.PP
> > > +In most cases,
> > > +.I len
> > > +is equal to
> > > +.I unencoded_len
> > > +and
> > > +.I unencoded_offset
> > > +is zero.
> > > +However, it may be necessary to refer to a subset of the unencoded data,
> > > +usually because a read occurred in the middle of an encoded extent,
> > > +because part of an extent was overwritten or deallocated in some
> > > +way (e.g., with
> > > +.BR write (2),
> > > +.BR truncate (2),
> > > +or
> > > +.BR fallocate (2))
> > > +or because part of an extent was added to the file (e.g., with
> > > +.BR ioctl_ficlonerange (2)
> > > +or
> > > +.BR ioctl_fideduperange (2)).
> > > +For example, if
> > > +.I len
> > > +is 300,
> > > +.I unencoded_len
> > > +is 1000,
> > > +and
> > > +.I unencoded_offset
> > > +is 600,
> > > +then the encoded data is 1000 bytes long when decoded,
> > > +of which only the 300 bytes starting at offset 600 are used;
> > > +the first 600 and last 100 bytes should be ignored.
> > > +.PP
> > > +If the unencoded data is actually longer than
> > > +.IR unencoded_len ,
> > > +then it is truncated;
> > > +if it is shorter, then it is extended with zeroes.
> >
> > I find the unencoded_len/unencoded_offset API extremely confusing and all
> > the clarifications above did not help to ease this feeling.
> > Please remind me why does the API need to expose unencoded details at all.
> > I understand the backup/restore use case for read/write encoded data.
> > I do not understand how unencoded offset info is relevant to this use case
> > or what are the other use cases it is relevant for.
>
> I agree, it's confusing. However, without this concept on the read side,
> there's no way to represent some file extent layouts, and without the
> write side, those layouts can't be written back out. That would make
> this interface much less useful for backups.
>
> These cases arise in a few ways on Btrfs:
>
> 1. Files with a size unaligned to the block size.
>
>    Ignoring inline data, Btrfs always pads data to the filesystem block
>    size when compressing. So, a file with a size unaligned to the block
>    size will end with an extent that decompresses to a multiple of the
>    block size, but logically the file only contains the data up to
>    i_size. In this case, len (length up to i_size) < unencoded_len (full
>    decompressed length). This can arise simply from writing out an
>    unaligned file or from truncating a file unaligned.
>
> 2. FICLONERANGE from the middle of an extent.
>
>    Suppose file A has a large compressed extent with
>    len = unencoded_len = 128k and unencoded_offset = 0. If the user does
>    an FICLONERANGE out of the middle of that extent (say, 64k long and
>    4k from the start of the extent), Btrfs creates a "partial" extent
>    which references the original extent (in my example, the result would
>    have len = 64k, unencoded_offset = 4k, and unencoded_len still 128k).
>
> 3. Overwriting the middle of an extent.
>
>    In some cases, when the middle of an extent is overwritten (e.g., an
>    O_DIRECT write, FICLONERANGE, or FIDEDUPERANGE), Btrfs splits up the
>    overwritten extents into partial extents referencing the original
>    extent instead of rewriting the whole extent.
>
> These aren't specific to compression or Btrfs' on-disk format. fscrypt
> uses block ciphers for file data, so case 1 is just as relevant for
> that. The way Btrfs handles case 2 is the only sane way I can see for
> supporting FICLONERANGE for encoded data.
>

I see... so now I understand the complication, but that doesn't mean
that the developers reading the encoded_io documentation will or that
they will get the implementation details right.

IMO, if the only use case for encoded io is backup/restore, then we
should make the API simpler and more oriented to this use case, namely,
serialization -
For all I care, btrfs can still return struct encoded_iov in iov[0],
but the user needs not to know about this and this internal detail should
not be documented nor exposed in UAPI.
btrfs send reads a stream of encoded data and metadata that describes it.
btrfs receive writes the encoded data stream and metadata descriptors that
tell the file system about overlapping extents and whatnot.

Is that something that can work out, or does userspace have to be aware
of encoded extents layout?

Thanks,
Amir.

Re: [PATCH man-pages v4] Document encoded I/O

[Omar Sandoval]

On Sun, Mar 01, 2020 at 09:26:10AM +0200, Amir Goldstein wrote:
> On Sat, Feb 29, 2020 at 8:03 PM Omar Sandoval <osandov@osandov.com> wrote:
> >
> > On Sat, Feb 29, 2020 at 12:28:41PM +0200, Amir Goldstein wrote:
> > > > +encoded_io \- overview of encoded I/O
> > > > +.SH DESCRIPTION
> > > > +Several filesystems (e.g., Btrfs) support transparent encoding
> > > > +(e.g., compression, encryption) of data on disk:
> > > > +written data is encoded by the kernel before it is written to disk,
> > > > +and read data is decoded before being returned to the user.
> > > > +In some cases, it is useful to skip this encoding step.
> > > > +For example, the user may want to read the compressed contents of a file
> > > > +or write pre-compressed data directly to a file.
> > > > +This is referred to as "encoded I/O".
> > > > +.SS Encoded I/O API
> > > > +Encoded I/O is specified with the
> > > > +.B RWF_ENCODED
> > > > +flag to
> > > > +.BR preadv2 (2)
> > > > +and
> > > > +.BR pwritev2 (2).
> > > > +If
> > > > +.B RWF_ENCODED
> > > > +is specified, then
> > > > +.I iov[0].iov_base
> > > > +points to an
> > > > +.I
> > > > +encoded_iov
> > > > +structure, defined in
> > > > +.I <linux/fs.h>
> > > > +as:
> > > > +.PP
> > > > +.in +4n
> > > > +.EX
> > > > +struct encoded_iov {
> > > > +    __aligned_u64 len;
> > > > +    __aligned_u64 unencoded_len;
> > > > +    __aligned_u64 unencoded_offset;
> > > > +    __u32 compression;
> > > > +    __u32 encryption;
> > > > +};
> > >
> > > This new API can generate many diverse error conditions that the standard errno
> > > codes are not rich enough to describe.
> > > Maybe add room for encoded io specific error codes in the metadata structure
> > > would be good practice, for example:
> > > - compression method not supported
> > > - encryption method not supported
> > > - the combination of enc/comp is not supported
> > > - and so on
> >
> > I like this idea, but it feels like even more iovec abuse. Namely, for
> 
> That's true.
> 
> > pwritev2(), it feels a little off that we'd be copying _to_ user memory
> > rather than only copying from. It's probably worth it for better errors,
> > though.
> >
> 
> Apropos iovec abuse, if encoded io is going to interpret iovec[0] differently
> why not interpret iovec arg differently. The result might be less awkward if
> the structure passed to preadv2/pwritev2 is struct encoded_iov * instead
> of struct iov *.

IMO, that's clunkier both from an API perspective and an implementation
perspective. On the implementation side, we now have to special case a
bunch of places in the VFS that are expecting a struct iovec *. On the
API side, it's so far from p{read,write}v2 that it might as well be an
ioctl or a new system call. (In fact, v1 of this series was a
Btrfs-specific ioctl, but it's much so nicer to reuse the VFS read/write
infrastructure.)

[snip]

> > > I find the unencoded_len/unencoded_offset API extremely confusing and all
> > > the clarifications above did not help to ease this feeling.
> > > Please remind me why does the API need to expose unencoded details at all.
> > > I understand the backup/restore use case for read/write encoded data.
> > > I do not understand how unencoded offset info is relevant to this use case
> > > or what are the other use cases it is relevant for.
> >
> > I agree, it's confusing. However, without this concept on the read side,
> > there's no way to represent some file extent layouts, and without the
> > write side, those layouts can't be written back out. That would make
> > this interface much less useful for backups.
> >
> > These cases arise in a few ways on Btrfs:
> >
> > 1. Files with a size unaligned to the block size.
> >
> >    Ignoring inline data, Btrfs always pads data to the filesystem block
> >    size when compressing. So, a file with a size unaligned to the block
> >    size will end with an extent that decompresses to a multiple of the
> >    block size, but logically the file only contains the data up to
> >    i_size. In this case, len (length up to i_size) < unencoded_len (full
> >    decompressed length). This can arise simply from writing out an
> >    unaligned file or from truncating a file unaligned.
> >
> > 2. FICLONERANGE from the middle of an extent.
> >
> >    Suppose file A has a large compressed extent with
> >    len = unencoded_len = 128k and unencoded_offset = 0. If the user does
> >    an FICLONERANGE out of the middle of that extent (say, 64k long and
> >    4k from the start of the extent), Btrfs creates a "partial" extent
> >    which references the original extent (in my example, the result would
> >    have len = 64k, unencoded_offset = 4k, and unencoded_len still 128k).
> >
> > 3. Overwriting the middle of an extent.
> >
> >    In some cases, when the middle of an extent is overwritten (e.g., an
> >    O_DIRECT write, FICLONERANGE, or FIDEDUPERANGE), Btrfs splits up the
> >    overwritten extents into partial extents referencing the original
> >    extent instead of rewriting the whole extent.
> >
> > These aren't specific to compression or Btrfs' on-disk format. fscrypt
> > uses block ciphers for file data, so case 1 is just as relevant for
> > that. The way Btrfs handles case 2 is the only sane way I can see for
> > supporting FICLONERANGE for encoded data.
> >
> 
> I see... so now I understand the complication, but that doesn't mean
> that the developers reading the encoded_io documentation will or that
> they will get the implementation details right.
> 
> IMO, if the only use case for encoded io is backup/restore, then we
> should make the API simpler and more oriented to this use case, namely,
> serialization -
> For all I care, btrfs can still return struct encoded_iov in iov[0],
> but the user needs not to know about this and this internal detail should
> not be documented nor exposed in UAPI.
> btrfs send reads a stream of encoded data and metadata that describes it.
> btrfs receive writes the encoded data stream and metadata descriptors that
> tell the file system about overlapping extents and whatnot.
> 
> Is that something that can work out, or does userspace have to be aware
> of encoded extents layout?

There are use cases outside of backups that would benefit from being
able to make arbitrary encoded writes. Specifically, one of my
colleagues at Facebook expressed interest in using encoded writes for
package distribution. The idea is that a package could be distributed as
a compressed archive and installed via encoded writes and reflinks to
the proper files, avoiding any need to decompress the package contents
before they're actually accessed. This sort of low-level fiddling needs
a proper UAPI. I'd much rather improve the documentation than make it
opaque.

Thanks,
Omar

Re: [PATCH man-pages v4] Document encoded I/O

[Michael Kerrisk (man-pages)]

Hello Omar,

(Unless you CC both me and mtk.manpages@gmail.com, it's easily
possible that I will miss your man-pages patches.)

What's the status here? I presume the features documented here are not
yet merged, right? Is the aim still to have them merged in the future?

Thanks,

Michael

On Sat, 29 Feb 2020 at 00:16, Omar Sandoval <osandov@osandov.com> wrote:
>
> From: Omar Sandoval <osandov@fb.com>
>
> This adds a new page, encoded_io(7), providing an overview of encoded
> I/O and updates fcntl(2), open(2), and preadv2(2)/pwritev2(2) to
> reference it.
>
> Signed-off-by: Omar Sandoval <osandov@fb.com>
> ---
>  man2/fcntl.2      |  10 +-
>  man2/open.2       |  13 ++
>  man2/readv.2      |  64 ++++++++++
>  man7/encoded_io.7 | 298 ++++++++++++++++++++++++++++++++++++++++++++++
>  4 files changed, 384 insertions(+), 1 deletion(-)
>  create mode 100644 man7/encoded_io.7
>
> diff --git a/man2/fcntl.2 b/man2/fcntl.2
> index bb1ac1f5d..15a1010a6 100644
> --- a/man2/fcntl.2
> +++ b/man2/fcntl.2
> @@ -222,8 +222,9 @@ On Linux, this command can change only the
>  .BR O_ASYNC ,
>  .BR O_DIRECT ,
>  .BR O_NOATIME ,
> +.BR O_NONBLOCK ,
>  and
> -.B O_NONBLOCK
> +.B O_ALLOW_ENCODED
>  flags.
>  It is not possible to change the
>  .BR O_DSYNC
> @@ -1821,6 +1822,13 @@ Attempted to clear the
>  flag on a file that has the append-only attribute set.
>  .TP
>  .B EPERM
> +Attempted to set the
> +.B O_ALLOW_ENCODED
> +flag and the calling process did not have the
> +.B CAP_SYS_ADMIN
> +capability.
> +.TP
> +.B EPERM
>  .I cmd
>  was
>  .BR F_ADD_SEALS ,
> diff --git a/man2/open.2 b/man2/open.2
> index 3ab4ee17b..256cb4247 100644
> --- a/man2/open.2
> +++ b/man2/open.2
> @@ -421,6 +421,14 @@ was followed by a call to
>  .BR fdatasync (2)).
>  .IR "See NOTES below" .
>  .TP
> +.B O_ALLOW_ENCODED
> +Open the file with encoded I/O permissions;
> +see
> +.BR encoded_io (7).
> +The caller must have the
> +.B CAP_SYS_ADMIN
> +capability.
> +.TP
>  .B O_EXCL
>  Ensure that this call creates the file:
>  if this flag is specified in conjunction with
> @@ -1176,6 +1184,11 @@ did not match the owner of the file and the caller was not privileged.
>  The operation was prevented by a file seal; see
>  .BR fcntl (2).
>  .TP
> +.B EPERM
> +The
> +.B O_ALLOW_ENCODED
> +flag was specified, but the caller was not privileged.
> +.TP
>  .B EROFS
>  .I pathname
>  refers to a file on a read-only filesystem and write access was
> diff --git a/man2/readv.2 b/man2/readv.2
> index af27aa63e..8b5458023 100644
> --- a/man2/readv.2
> +++ b/man2/readv.2
> @@ -265,6 +265,11 @@ the data is always appended to the end of the file.
>  However, if the
>  .I offset
>  argument is \-1, the current file offset is updated.
> +.TP
> +.BR RWF_ENCODED " (since Linux 5.7)"
> +Read or write encoded (e.g., compressed) data.
> +See
> +.BR encoded_io (7).
>  .SH RETURN VALUE
>  On success,
>  .BR readv (),
> @@ -284,6 +289,13 @@ than requested (see
>  and
>  .BR write (2)).
>  .PP
> +If
> +.B
> +RWF_ENCODED
> +was specified in
> +.IR flags ,
> +then the return value is the number of encoded bytes.
> +.PP
>  On error, \-1 is returned, and \fIerrno\fP is set appropriately.
>  .SH ERRORS
>  The errors are as given for
> @@ -314,6 +326,58 @@ is less than zero or greater than the permitted maximum.
>  .TP
>  .B EOPNOTSUPP
>  An unknown flag is specified in \fIflags\fP.
> +.TP
> +.B EOPNOTSUPP
> +.B RWF_ENCODED
> +is specified in
> +.I flags
> +and the filesystem does not implement encoded I/O.
> +.TP
> +.B EPERM
> +.B RWF_ENCODED
> +is specified in
> +.I flags
> +and the file was not opened with the
> +.B O_ALLOW_ENCODED
> +flag.
> +.PP
> +.BR preadv2 ()
> +can fail for the following reasons:
> +.TP
> +.B E2BIG
> +.B RWF_ENCODED
> +is specified in
> +.I flags
> +and
> +.I iov[0]
> +is not large enough to return the encoding metadata.
> +.TP
> +.B ENOBUFS
> +.B RWF_ENCODED
> +is specified in
> +.I flags
> +and the buffers in
> +.I iov
> +are not big enough to return the encoded data.
> +.PP
> +.BR pwritev2 ()
> +can fail for the following reasons:
> +.TP
> +.B E2BIG
> +.B RWF_ENCODED
> +is specified in
> +.I flags
> +and
> +.I iov[0]
> +contains non-zero fields
> +after the kernel's
> +.IR "sizeof(struct\ encoded_iov)" .
> +.TP
> +.B EINVAL
> +.B RWF_ENCODED
> +is specified in
> +.I flags
> +and the alignment and/or size requirements are not met.
>  .SH VERSIONS
>  .BR preadv ()
>  and
> diff --git a/man7/encoded_io.7 b/man7/encoded_io.7
> new file mode 100644
> index 000000000..72b40353f
> --- /dev/null
> +++ b/man7/encoded_io.7
> @@ -0,0 +1,298 @@
> +.\" Copyright (c) 2019 by Omar Sandoval <osandov@fb.com>
> +.\"
> +.\" %%%LICENSE_START(VERBATIM)
> +.\" Permission is granted to make and distribute verbatim copies of this
> +.\" manual provided the copyright notice and this permission notice are
> +.\" preserved on all copies.
> +.\"
> +.\" Permission is granted to copy and distribute modified versions of this
> +.\" manual under the conditions for verbatim copying, provided that the
> +.\" entire resulting derived work is distributed under the terms of a
> +.\" permission notice identical to this one.
> +.\"
> +.\" Since the Linux kernel and libraries are constantly changing, this
> +.\" manual page may be incorrect or out-of-date.  The author(s) assume no
> +.\" responsibility for errors or omissions, or for damages resulting from
> +.\" the use of the information contained herein.  The author(s) may not
> +.\" have taken the same level of care in the production of this manual,
> +.\" which is licensed free of charge, as they might when working
> +.\" professionally.
> +.\"
> +.\" Formatted or processed versions of this manual, if unaccompanied by
> +.\" the source, must acknowledge the copyright and authors of this work.
> +.\" %%%LICENSE_END
> +.\"
> +.\"
> +.TH ENCODED_IO  7 2019-10-14 "Linux" "Linux Programmer's Manual"
> +.SH NAME
> +encoded_io \- overview of encoded I/O
> +.SH DESCRIPTION
> +Several filesystems (e.g., Btrfs) support transparent encoding
> +(e.g., compression, encryption) of data on disk:
> +written data is encoded by the kernel before it is written to disk,
> +and read data is decoded before being returned to the user.
> +In some cases, it is useful to skip this encoding step.
> +For example, the user may want to read the compressed contents of a file
> +or write pre-compressed data directly to a file.
> +This is referred to as "encoded I/O".
> +.SS Encoded I/O API
> +Encoded I/O is specified with the
> +.B RWF_ENCODED
> +flag to
> +.BR preadv2 (2)
> +and
> +.BR pwritev2 (2).
> +If
> +.B RWF_ENCODED
> +is specified, then
> +.I iov[0].iov_base
> +points to an
> +.I
> +encoded_iov
> +structure, defined in
> +.I <linux/fs.h>
> +as:
> +.PP
> +.in +4n
> +.EX
> +struct encoded_iov {
> +    __aligned_u64 len;
> +    __aligned_u64 unencoded_len;
> +    __aligned_u64 unencoded_offset;
> +    __u32 compression;
> +    __u32 encryption;
> +};
> +.EE
> +.in
> +.PP
> +This may be extended in the future, so
> +.I iov[0].iov_len
> +must be set to
> +.I "sizeof(struct\ encoded_iov)"
> +for forward/backward compatibility.
> +The remaining buffers contain the encoded data.
> +.PP
> +.I compression
> +and
> +.I encryption
> +are the encoding fields.
> +.I compression
> +is one of
> +.B ENCODED_IOV_COMPRESSION_NONE
> +(zero),
> +.BR ENCODED_IOV_COMPRESSION_ZLIB ,
> +.BR ENCODED_IOV_COMPRESSION_LZO ,
> +or
> +.BR ENCODED_IOV_COMPRESSION_ZSTD .
> +.I encryption
> +is currently always
> +.B ENCODED_IOV_ENCRYPTION_NONE
> +(zero).
> +.PP
> +.I unencoded_len
> +is the length of the unencoded (i.e., decrypted and decompressed) data.
> +.I unencoded_offset
> +is the offset into the unencoded data where the data in the file begins
> +(less than or equal to
> +.IR unencoded_len ).
> +.I len
> +is the length of the data in the file
> +(less than or equal to
> +.I unencoded_len
> +-
> +.IR unencoded_offset ).
> +.I
> +.PP
> +In most cases,
> +.I len
> +is equal to
> +.I unencoded_len
> +and
> +.I unencoded_offset
> +is zero.
> +However, it may be necessary to refer to a subset of the unencoded data,
> +usually because a read occurred in the middle of an encoded extent,
> +because part of an extent was overwritten or deallocated in some
> +way (e.g., with
> +.BR write (2),
> +.BR truncate (2),
> +or
> +.BR fallocate (2))
> +or because part of an extent was added to the file (e.g., with
> +.BR ioctl_ficlonerange (2)
> +or
> +.BR ioctl_fideduperange (2)).
> +For example, if
> +.I len
> +is 300,
> +.I unencoded_len
> +is 1000,
> +and
> +.I unencoded_offset
> +is 600,
> +then the encoded data is 1000 bytes long when decoded,
> +of which only the 300 bytes starting at offset 600 are used;
> +the first 600 and last 100 bytes should be ignored.
> +.PP
> +If the unencoded data is actually longer than
> +.IR unencoded_len ,
> +then it is truncated;
> +if it is shorter, then it is extended with zeroes.
> +.PP
> +For
> +.BR pwritev2 (),
> +the metadata should be specified in
> +.IR iov[0] .
> +If
> +.I iov[0].iov_len
> +is less than
> +.I "sizeof(struct\ encoded_iov)"
> +in the kernel,
> +then any fields unknown to userspace are treated as if they were zero;
> +if it is greater and any fields unknown to the kernel are non-zero,
> +then this returns -1 and sets
> +.I errno
> +to
> +.BR E2BIG .
> +The encoded data should be passed in the remaining buffers.
> +This returns the number of encoded bytes written (that is, the sum of
> +.I iov[n].iov_len
> +for 1 <=
> +.I n
> +<
> +.IR iovcnt ;
> +partial writes will not occur).
> +If the
> +.I offset
> +argument to
> +.BR pwritev2 ()
> +is -1, then the file offset is incremented by
> +.IR len .
> +At least one encoding field must be non-zero.
> +Note that the encoded data is not validated when it is written;
> +if it is not valid (e.g., it cannot be decompressed),
> +then a subsequent read may return an error.
> +.PP
> +For
> +.BR preadv2 (),
> +the metadata is returned in
> +.IR iov[0] .
> +If
> +.I iov[0].iov_len
> +is less than
> +.I "sizeof(struct\ encoded_iov)"
> +in the kernel and any fields unknown to userspace are non-zero,
> +then this returns -1 and sets
> +.I errno
> +to
> +.BR E2BIG ;
> +if it is greater,
> +then any fields unknown to the kernel are returned as zero.
> +The encoded data is returned in the remaining buffers.
> +If the provided buffers are not large enough to return an entire encoded
> +extent,
> +then this returns -1 and sets
> +.I errno
> +to
> +.BR ENOBUFS .
> +This returns the number of encoded bytes read.
> +If the
> +.I offset
> +argument to
> +.BR preadv2 ()
> +is -1, then the file offset is incremented by
> +.IR len .
> +This will only return one encoded extent per call.
> +This can also read data which is not encoded;
> +all encoding fields will be zero in that case.
> +.PP
> +As the filesystem page cache typically contains decoded data,
> +encoded I/O bypasses the page cache.
> +.SS Security
> +Encoded I/O creates the potential for some security issues:
> +.IP * 3
> +Encoded writes allow writing arbitrary data which the kernel will decode on
> +a subsequent read. Decompression algorithms are complex and may have bugs
> +which can be exploited by maliciously crafted data.
> +.IP *
> +Encoded reads may return data which is not logically present in the file
> +(see the discussion of
> +.I len
> +vs.
> +.I unencoded_len
> +above).
> +It may not be intended for this data to be readable.
> +.PP
> +Therefore, encoded I/O requires privilege.
> +Namely, the
> +.B RWF_ENCODED
> +flag may only be used when the file was opened with the
> +.B O_ALLOW_ENCODED
> +flag to
> +.BR open (2),
> +which requires the
> +.B CAP_SYS_ADMIN
> +capability.
> +.B O_ALLOW_ENCODED
> +may be set and cleared with
> +.BR fcntl (2).
> +Note that it is not cleared on
> +.BR fork (2)
> +or
> +.BR execve (2);
> +one may wish to use
> +.B O_CLOEXEC
> +with
> +.BR O_ALLOW_ENCODED .
> +.SS Filesystem support
> +Encoded I/O is supported on the following filesystems:
> +.TP
> +Btrfs (since Linux 5.8)
> +.IP
> +Btrfs supports encoded reads and writes of compressed data.
> +The data is encoded as follows:
> +.RS
> +.IP * 3
> +If
> +.I compression
> +is
> +.BR ENCODED_IOV_COMPRESSION_ZLIB ,
> +then the encoded data is a single zlib stream.
> +.IP *
> +If
> +.I compression
> +is
> +.BR ENCODED_IOV_COMPRESSION_LZO ,
> +then the encoded data is compressed page by page with LZO1X
> +and wrapped in the format documented in the Linux kernel source file
> +.IR fs/btrfs/lzo.c .
> +.IP *
> +If
> +.I compression
> +is
> +.BR ENCODED_IOV_COMPRESSION_ZSTD ,
> +then the encoded data is a single zstd frame compressed with the
> +.I windowLog
> +compression parameter set to no more than 17.
> +.RE
> +.IP
> +Additionally, there are some restrictions on
> +.BR pwritev2 ():
> +.RS
> +.IP * 3
> +.I offset
> +(or the current file offset if
> +.I offset
> +is -1) must be aligned to the sector size of the filesystem.
> +.IP *
> +.I len
> +must be aligned to the sector size of the filesystem
> +unless the data ends at or beyond the current end of the file.
> +.IP *
> +.I unencoded_len
> +and the length of the encoded data must each be no more than 128 KiB.
> +This limit may increase in the future.
> +.IP *
> +The length of the encoded data must be less than or equal to
> +.IR unencoded_len .
> +.RE
> --
> 2.25.1
>


-- 
Michael Kerrisk
Linux man-pages maintainer; http://www.kernel.org/doc/man-pages/
Linux/UNIX System Programming Training: http://man7.org/training/

Re: [PATCH man-pages v4] Document encoded I/O

[Omar Sandoval]

On Thu, Apr 16, 2020 at 02:26:01PM +0200, Michael Kerrisk (man-pages) wrote:
> Hello Omar,
> 
> (Unless you CC both me and mtk.manpages@gmail.com, it's easily
> possible that I will miss your man-pages patches.)

That's good to know, thanks. Do you mind being CCd on man-pages for
features that haven't been finalized yet?

> What's the status here? I presume the features documented here are not
> yet merged, right? Is the aim still to have them merged in the future?

They're not yet merged but I'm still working on having them merged. I'm
still waiting for VFS review.

Thanks!

Re: [PATCH man-pages v4] Document encoded I/O

[Michael Kerrisk (man-pages)]

Hello Omar,

On Thu, 16 Apr 2020 at 19:02, Omar Sandoval <osandov@osandov.com> wrote:
>
> On Thu, Apr 16, 2020 at 02:26:01PM +0200, Michael Kerrisk (man-pages) wrote:
> > Hello Omar,
> >
> > (Unless you CC both me and mtk.manpages@gmail.com, it's easily
> > possible that I will miss your man-pages patches.)
>
> That's good to know, thanks. Do you mind being CCd on man-pages for
> features that haven't been finalized yet?

Please do CC me and linux-man@ on such patches. Just make sure that
the patch notes that the feature is not yet upstream.
>
> > What's the status here? I presume the features documented here are not
> > yet merged, right? Is the aim still to have them merged in the future?
>
> They're not yet merged but I'm still working on having them merged. I'm
> still waiting for VFS review.

Okay.

Thanks,

Michael

-- 
Michael Kerrisk
Linux man-pages maintainer; http://www.kernel.org/doc/man-pages/
Linux/UNIX System Programming Training: http://man7.org/training/

[PATCH v4 1/9] iov_iter: add copy_struct_from_iter()

[Omar Sandoval]

From: Omar Sandoval <osandov@fb.com>

This is essentially copy_struct_from_user() but for an iov_iter.

Suggested-by: Aleksa Sarai <cyphar@cyphar.com>
Signed-off-by: Omar Sandoval <osandov@fb.com>
---
 include/linux/uio.h |  2 ++
 lib/iov_iter.c      | 82 +++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 84 insertions(+)

diff --git a/include/linux/uio.h b/include/linux/uio.h
index 9576fd8158d7..9776b4be4833 100644
--- a/include/linux/uio.h
+++ b/include/linux/uio.h
@@ -122,6 +122,8 @@ size_t copy_page_to_iter(struct page *page, size_t offset, size_t bytes,
 			 struct iov_iter *i);
 size_t copy_page_from_iter(struct page *page, size_t offset, size_t bytes,
 			 struct iov_iter *i);
+int copy_struct_from_iter(void *dst, size_t ksize, struct iov_iter *i,
+			  size_t usize);
 
 size_t _copy_to_iter(const void *addr, size_t bytes, struct iov_iter *i);
 size_t _copy_from_iter(void *addr, size_t bytes, struct iov_iter *i);
diff --git a/lib/iov_iter.c b/lib/iov_iter.c
index 51595bf3af85..a2a6ab41767b 100644
--- a/lib/iov_iter.c
+++ b/lib/iov_iter.c
@@ -942,6 +942,88 @@ size_t copy_page_from_iter(struct page *page, size_t offset, size_t bytes,
 }
 EXPORT_SYMBOL(copy_page_from_iter);
 
+/**
+ * copy_struct_from_iter - copy a struct from an iov_iter
+ * @dst: Destination buffer.
+ * @ksize: Size of @dst struct.
+ * @i: Source iterator.
+ * @usize: (Alleged) size of struct in @i.
+ *
+ * Copies a struct from an iov_iter in a way that guarantees
+ * backwards-compatibility for struct arguments in an iovec (as long as the
+ * rules for copy_struct_from_user() are followed).
+ *
+ * The recommended usage is that @usize be taken from the current segment:
+ *
+ *   int do_foo(struct iov_iter *i)
+ *   {
+ *     size_t usize = iov_iter_single_seg_count(i);
+ *     struct foo karg;
+ *     int err;
+ *
+ *     if (usize > PAGE_SIZE)
+ *       return -E2BIG;
+ *     if (usize < FOO_SIZE_VER0)
+ *       return -EINVAL;
+ *     err = copy_struct_from_iter(&karg, sizeof(karg), i, usize);
+ *     if (err)
+ *       return err;
+ *
+ *     // ...
+ *   }
+ *
+ * Return: 0 on success, -errno on error (see copy_struct_from_user()).
+ *
+ * On success, the iterator is advanced @usize bytes. On error, the iterator is
+ * not advanced.
+ */
+int copy_struct_from_iter(void *dst, size_t ksize, struct iov_iter *i,
+			  size_t usize)
+{
+	if (usize <= ksize) {
+		if (!copy_from_iter_full(dst, usize, i))
+			return -EFAULT;
+		memset(dst + usize, 0, ksize - usize);
+	} else {
+		size_t copied = 0, copy;
+		int ret;
+
+		if (WARN_ON(iov_iter_is_pipe(i)) || unlikely(i->count < usize))
+			return -EFAULT;
+		if (iter_is_iovec(i))
+			might_fault();
+		iterate_all_kinds(i, usize, v, ({
+			copy = min(ksize - copied, v.iov_len);
+			if (copy && copyin(dst + copied, v.iov_base, copy))
+				return -EFAULT;
+			copied += copy;
+			ret = check_zeroed_user(v.iov_base + copy,
+						v.iov_len - copy);
+			if (ret <= 0)
+				return ret ?: -E2BIG;
+			0;}), ({
+			char *addr = kmap_atomic(v.bv_page);
+			copy = min_t(size_t, ksize - copied, v.bv_len);
+			memcpy(dst + copied, addr + v.bv_offset, copy);
+			copied += copy;
+			ret = memchr_inv(addr + v.bv_offset + copy, 0,
+					 v.bv_len - copy) ? -E2BIG : 0;
+			kunmap_atomic(addr);
+			if (ret)
+				return ret;
+			}), ({
+			copy = min(ksize - copied, v.iov_len);
+			memcpy(dst + copied, v.iov_base, copy);
+			if (memchr_inv(v.iov_base, 0, v.iov_len))
+				return -E2BIG;
+			})
+		)
+		iov_iter_advance(i, usize);
+	}
+	return 0;
+}
+EXPORT_SYMBOL_GPL(copy_struct_from_iter);
+
 static size_t pipe_zero(size_t bytes, struct iov_iter *i)
 {
 	struct pipe_inode_info *pipe = i->pipe;
-- 
2.25.1

[PATCH v4 2/9] fs: add O_ALLOW_ENCODED open flag

[Omar Sandoval]

From: Omar Sandoval <osandov@fb.com>

The upcoming RWF_ENCODED operation introduces some security concerns:

1. Compressed writes will pass arbitrary data to decompression
   algorithms in the kernel.
2. Compressed reads can leak truncated/hole punched data.

Therefore, we need to require privilege for RWF_ENCODED. It's not
possible to do the permissions checks at the time of the read or write
because, e.g., io_uring submits IO from a worker thread. So, add an open
flag which requires CAP_SYS_ADMIN. It can also be set and cleared with
fcntl(). The flag is not cleared in any way on fork or exec; it should
probably be used with O_CLOEXEC in most cases.

Note that the usual issue that unknown open flags are ignored doesn't
really matter for O_ALLOW_ENCODED; if the kernel doesn't support
O_ALLOW_ENCODED, then it doesn't support RWF_ENCODED, either.

Signed-off-by: Omar Sandoval <osandov@fb.com>
---
 arch/alpha/include/uapi/asm/fcntl.h  |  1 +
 arch/parisc/include/uapi/asm/fcntl.h |  1 +
 arch/sparc/include/uapi/asm/fcntl.h  |  1 +
 fs/fcntl.c                           | 10 ++++++++--
 fs/namei.c                           |  4 ++++
 include/linux/fcntl.h                |  2 +-
 include/uapi/asm-generic/fcntl.h     |  4 ++++
 7 files changed, 20 insertions(+), 3 deletions(-)

diff --git a/arch/alpha/include/uapi/asm/fcntl.h b/arch/alpha/include/uapi/asm/fcntl.h
index 50bdc8e8a271..391e0d112e41 100644
--- a/arch/alpha/include/uapi/asm/fcntl.h
+++ b/arch/alpha/include/uapi/asm/fcntl.h
@@ -34,6 +34,7 @@
 
 #define O_PATH		040000000
 #define __O_TMPFILE	0100000000
+#define O_ALLOW_ENCODED	0200000000
 
 #define F_GETLK		7
 #define F_SETLK		8
diff --git a/arch/parisc/include/uapi/asm/fcntl.h b/arch/parisc/include/uapi/asm/fcntl.h
index 03ce20e5ad7d..1188b27002b3 100644
--- a/arch/parisc/include/uapi/asm/fcntl.h
+++ b/arch/parisc/include/uapi/asm/fcntl.h
@@ -22,6 +22,7 @@
 
 #define O_PATH		020000000
 #define __O_TMPFILE	040000000
+#define O_ALLOW_ENCODED	100000000
 
 #define F_GETLK64	8
 #define F_SETLK64	9
diff --git a/arch/sparc/include/uapi/asm/fcntl.h b/arch/sparc/include/uapi/asm/fcntl.h
index 67dae75e5274..ac3e8c9cb32c 100644
--- a/arch/sparc/include/uapi/asm/fcntl.h
+++ b/arch/sparc/include/uapi/asm/fcntl.h
@@ -37,6 +37,7 @@
 
 #define O_PATH		0x1000000
 #define __O_TMPFILE	0x2000000
+#define O_ALLOW_ENCODED	0x8000000
 
 #define F_GETOWN	5	/*  for sockets. */
 #define F_SETOWN	6	/*  for sockets. */
diff --git a/fs/fcntl.c b/fs/fcntl.c
index 9bc167562ee8..c5e169115e55 100644
--- a/fs/fcntl.c
+++ b/fs/fcntl.c
@@ -30,7 +30,8 @@
 #include <asm/siginfo.h>
 #include <linux/uaccess.h>
 
-#define SETFL_MASK (O_APPEND | O_NONBLOCK | O_NDELAY | O_DIRECT | O_NOATIME)
+#define SETFL_MASK (O_APPEND | O_NONBLOCK | O_NDELAY | O_DIRECT | O_NOATIME | \
+		    O_ALLOW_ENCODED)
 
 static int setfl(int fd, struct file * filp, unsigned long arg)
 {
@@ -49,6 +50,11 @@ static int setfl(int fd, struct file * filp, unsigned long arg)
 		if (!inode_owner_or_capable(inode))
 			return -EPERM;
 
+	/* O_ALLOW_ENCODED can only be set by superuser */
+	if ((arg & O_ALLOW_ENCODED) && !(filp->f_flags & O_ALLOW_ENCODED) &&
+	    !capable(CAP_SYS_ADMIN))
+		return -EPERM;
+
 	/* required for strict SunOS emulation */
 	if (O_NONBLOCK != O_NDELAY)
 	       if (arg & O_NDELAY)
@@ -1031,7 +1037,7 @@ static int __init fcntl_init(void)
 	 * Exceptions: O_NONBLOCK is a two bit define on parisc; O_NDELAY
 	 * is defined as O_NONBLOCK on some platforms and not on others.
 	 */
-	BUILD_BUG_ON(21 - 1 /* for O_RDONLY being 0 */ !=
+	BUILD_BUG_ON(22 - 1 /* for O_RDONLY being 0 */ !=
 		HWEIGHT32(
 			(VALID_OPEN_FLAGS & ~(O_NONBLOCK | O_NDELAY)) |
 			__FMODE_EXEC | __FMODE_NONOTIFY));
diff --git a/fs/namei.c b/fs/namei.c
index db6565c99825..4546de90281a 100644
--- a/fs/namei.c
+++ b/fs/namei.c
@@ -3063,6 +3063,10 @@ static int may_open(const struct path *path, int acc_mode, int flag)
 	if (flag & O_NOATIME && !inode_owner_or_capable(inode))
 		return -EPERM;
 
+	/* O_ALLOW_ENCODED can only be set by superuser */
+	if ((flag & O_ALLOW_ENCODED) && !capable(CAP_SYS_ADMIN))
+		return -EPERM;
+
 	return 0;
 }
 
diff --git a/include/linux/fcntl.h b/include/linux/fcntl.h
index 7bcdcf4f6ab2..670939ea3c80 100644
--- a/include/linux/fcntl.h
+++ b/include/linux/fcntl.h
@@ -10,7 +10,7 @@
 	(O_RDONLY | O_WRONLY | O_RDWR | O_CREAT | O_EXCL | O_NOCTTY | O_TRUNC | \
 	 O_APPEND | O_NDELAY | O_NONBLOCK | O_NDELAY | __O_SYNC | O_DSYNC | \
 	 FASYNC	| O_DIRECT | O_LARGEFILE | O_DIRECTORY | O_NOFOLLOW | \
-	 O_NOATIME | O_CLOEXEC | O_PATH | __O_TMPFILE)
+	 O_NOATIME | O_CLOEXEC | O_PATH | __O_TMPFILE | O_ALLOW_ENCODED)
 
 /* List of all valid flags for the how->upgrade_mask argument: */
 #define VALID_UPGRADE_FLAGS \
diff --git a/include/uapi/asm-generic/fcntl.h b/include/uapi/asm-generic/fcntl.h
index 9dc0bf0c5a6e..75321c7a66ac 100644
--- a/include/uapi/asm-generic/fcntl.h
+++ b/include/uapi/asm-generic/fcntl.h
@@ -89,6 +89,10 @@
 #define __O_TMPFILE	020000000
 #endif
 
+#ifndef O_ALLOW_ENCODED
+#define O_ALLOW_ENCODED	040000000
+#endif
+
 /* a horrid kludge trying to make sure that this will fail on old kernels */
 #define O_TMPFILE (__O_TMPFILE | O_DIRECTORY)
 #define O_TMPFILE_MASK (__O_TMPFILE | O_DIRECTORY | O_CREAT)      
-- 
2.25.1

Re: [PATCH v4 2/9] fs: add O_ALLOW_ENCODED open flag

[Amir Goldstein]

On Sat, Feb 29, 2020 at 1:14 AM Omar Sandoval <osandov@osandov.com> wrote:
>
> From: Omar Sandoval <osandov@fb.com>
>
> The upcoming RWF_ENCODED operation introduces some security concerns:
>
> 1. Compressed writes will pass arbitrary data to decompression
>    algorithms in the kernel.
> 2. Compressed reads can leak truncated/hole punched data.
>
> Therefore, we need to require privilege for RWF_ENCODED. It's not
> possible to do the permissions checks at the time of the read or write
> because, e.g., io_uring submits IO from a worker thread. So, add an open
> flag which requires CAP_SYS_ADMIN. It can also be set and cleared with
> fcntl(). The flag is not cleared in any way on fork or exec; it should
> probably be used with O_CLOEXEC in most cases.
>

So let's be more proactive and disallow setting O_ALLOW_ENCODED without
O_CLOEXEC, shall we?

> Note that the usual issue that unknown open flags are ignored doesn't
> really matter for O_ALLOW_ENCODED; if the kernel doesn't support
> O_ALLOW_ENCODED, then it doesn't support RWF_ENCODED, either.
>

And for that matter, setting O_ALLOW_ENCODED without O_CLOEXEC
won't do any harm with old kernels - even better, it can serve as a fast
test for kernel RWF_ENCODED support using only the openat() syscall.

Thanks,
Amir.

[PATCH v4 3/9] fs: add RWF_ENCODED for reading/writing compressed data

[Omar Sandoval]

From: Omar Sandoval <osandov@fb.com>

Btrfs supports transparent compression: data written by the user can be
compressed when written to disk and decompressed when read back.
However, we'd like to add an interface to write pre-compressed data
directly to the filesystem, and the matching interface to read
compressed data without decompressing it. This adds support for
so-called "encoded I/O" via preadv2() and pwritev2().

A new RWF_ENCODED flags indicates that a read or write is "encoded". If
this flag is set, iov[0].iov_base points to a struct encoded_iov which
is used for metadata: namely, the compression algorithm, unencoded
(i.e., decompressed) length, and what subrange of the unencoded data
should be used (needed for truncated or hole-punched extents and when
reading in the middle of an extent). For reads, the filesystem returns
this information; for writes, the caller provides it to the filesystem.
iov[0].iov_len must be set to sizeof(struct encoded_iov), which can be
used to extend the interface in the future a la copy_struct_from_user().
The remaining iovecs contain the encoded extent.

This adds the VFS helpers for supporting encoded I/O and documentation
for filesystem support.

Signed-off-by: Omar Sandoval <osandov@fb.com>
---
 Documentation/filesystems/encoded_io.rst |  74 ++++++++++
 Documentation/filesystems/index.rst      |   1 +
 include/linux/fs.h                       |  16 +++
 include/uapi/linux/fs.h                  |  33 ++++-
 mm/filemap.c                             | 166 +++++++++++++++++++++--
 5 files changed, 276 insertions(+), 14 deletions(-)
 create mode 100644 Documentation/filesystems/encoded_io.rst

diff --git a/Documentation/filesystems/encoded_io.rst b/Documentation/filesystems/encoded_io.rst
new file mode 100644
index 000000000000..50405276d866
--- /dev/null
+++ b/Documentation/filesystems/encoded_io.rst
@@ -0,0 +1,74 @@
+===========
+Encoded I/O
+===========
+
+Encoded I/O is a mechanism for reading and writing encoded (e.g., compressed
+and/or encrypted) data directly from/to the filesystem. The userspace interface
+is thoroughly described in the :manpage:`encoded_io(7)` man page; this document
+describes the requirements for filesystem support.
+
+First of all, a filesystem supporting encoded I/O must indicate this by setting
+the ``FMODE_ENCODED_IO`` flag in its ``file_open`` file operation::
+
+    static int foo_file_open(struct inode *inode, struct file *filp)
+    {
+            ...
+            filep->f_mode |= FMODE_ENCODED_IO;
+            ...
+    }
+
+Encoded I/O goes through ``read_iter`` and ``write_iter``, designated by the
+``IOCB_ENCODED`` flag in ``kiocb->ki_flags``.
+
+Reads
+=====
+
+Encoded ``read_iter`` should:
+
+1. Call ``generic_encoded_read_checks()`` to validate the file and buffers
+   provided by userspace.
+2. Initialize the ``encoded_iov`` appropriately.
+3. Copy it to the user with ``copy_encoded_iov_to_iter()``.
+4. Copy the encoded data to the user.
+5. Advance ``kiocb->ki_pos`` by ``encoded_iov->len``.
+6. Return the size of the encoded data read, not including the ``encoded_iov``.
+
+There are a few details to be aware of:
+
+* Encoded ``read_iter`` should support reading unencoded data if the extent is
+  not encoded.
+* If the buffers provided by the user are not large enough to contain an entire
+  encoded extent, then ``read_iter`` should return ``-ENOBUFS``. This is to
+  avoid confusing userspace with truncated data that cannot be properly
+  decoded.
+* Reads in the middle of an encoded extent can be returned by setting
+  ``encoded_iov->unencoded_offset`` to non-zero.
+* Truncated unencoded data (e.g., because the file does not end on a block
+  boundary) may be returned by setting ``encoded_iov->len`` to a value smaller
+  value than ``encoded_iov->unencoded_len - encoded_iov->unencoded_offset``.
+
+Writes
+======
+
+Encoded ``write_iter`` should (in addition to the usual accounting/checks done
+by ``write_iter``):
+
+1. Call ``copy_encoded_iov_from_iter()`` to get and validate the
+   ``encoded_iov``.
+2. Call ``generic_encoded_write_checks()`` instead of
+   ``generic_write_checks()``.
+3. Check that the provided encoding in ``encoded_iov`` is supported.
+4. Advance ``kiocb->ki_pos`` by ``encoded_iov->len``.
+5. Return the size of the encoded data written.
+
+Again, there are a few details:
+
+* Encoded ``write_iter`` doesn't need to support writing unencoded data.
+* ``write_iter`` should either write all of the encoded data or none of it; it
+  must not do partial writes.
+* ``write_iter`` doesn't need to validate the encoded data; a subsequent read
+  may return, e.g., ``-EIO`` if the data is not valid.
+* The user may lie about the unencoded size of the data; a subsequent read
+  should truncate or zero-extend the unencoded data rather than returning an
+  error.
+* Be careful of page cache coherency.
diff --git a/Documentation/filesystems/index.rst b/Documentation/filesystems/index.rst
index 386eaad008b2..e074a3f1f856 100644
--- a/Documentation/filesystems/index.rst
+++ b/Documentation/filesystems/index.rst
@@ -37,6 +37,7 @@ filesystem implementations.
    journalling
    fscrypt
    fsverity
+   encoded_io
 
 Filesystems
 ===========
diff --git a/include/linux/fs.h b/include/linux/fs.h
index 3cd4fe6b845e..aa7efd3430d1 100644
--- a/include/linux/fs.h
+++ b/include/linux/fs.h
@@ -175,6 +175,9 @@ typedef int (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
 /* File does not contribute to nr_files count */
 #define FMODE_NOACCOUNT		((__force fmode_t)0x20000000)
 
+/* File supports encoded IO */
+#define FMODE_ENCODED_IO	((__force fmode_t)0x40000000)
+
 /*
  * Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector
  * that indicates that they should check the contents of the iovec are
@@ -314,6 +317,7 @@ enum rw_hint {
 #define IOCB_SYNC		(1 << 5)
 #define IOCB_WRITE		(1 << 6)
 #define IOCB_NOWAIT		(1 << 7)
+#define IOCB_ENCODED		(1 << 8)
 
 struct kiocb {
 	struct file		*ki_filp;
@@ -3109,6 +3113,13 @@ extern int sb_min_blocksize(struct super_block *, int);
 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
 extern ssize_t generic_write_checks(struct kiocb *, struct iov_iter *);
+struct encoded_iov;
+extern int generic_encoded_write_checks(struct kiocb *,
+					const struct encoded_iov *);
+extern int copy_encoded_iov_from_iter(struct encoded_iov *, struct iov_iter *);
+extern ssize_t generic_encoded_read_checks(struct kiocb *, struct iov_iter *);
+extern int copy_encoded_iov_to_iter(const struct encoded_iov *,
+				    struct iov_iter *);
 extern int generic_remap_checks(struct file *file_in, loff_t pos_in,
 				struct file *file_out, loff_t pos_out,
 				loff_t *count, unsigned int remap_flags);
@@ -3434,6 +3445,11 @@ static inline int kiocb_set_rw_flags(struct kiocb *ki, rwf_t flags)
 			return -EOPNOTSUPP;
 		ki->ki_flags |= IOCB_NOWAIT;
 	}
+	if (flags & RWF_ENCODED) {
+		if (!(ki->ki_filp->f_mode & FMODE_ENCODED_IO))
+			return -EOPNOTSUPP;
+		ki->ki_flags |= IOCB_ENCODED;
+	}
 	if (flags & RWF_HIPRI)
 		ki->ki_flags |= IOCB_HIPRI;
 	if (flags & RWF_DSYNC)
diff --git a/include/uapi/linux/fs.h b/include/uapi/linux/fs.h
index 379a612f8f1d..f8c6c1e08def 100644
--- a/include/uapi/linux/fs.h
+++ b/include/uapi/linux/fs.h
@@ -278,6 +278,34 @@ struct fsxattr {
 					 SYNC_FILE_RANGE_WAIT_BEFORE | \
 					 SYNC_FILE_RANGE_WAIT_AFTER)
 
+enum {
+	ENCODED_IOV_COMPRESSION_NONE,
+#define ENCODED_IOV_COMPRESSION_NONE ENCODED_IOV_COMPRESSION_NONE
+	ENCODED_IOV_COMPRESSION_ZLIB,
+#define ENCODED_IOV_COMPRESSION_ZLIB ENCODED_IOV_COMPRESSION_ZLIB
+	ENCODED_IOV_COMPRESSION_LZO,
+#define ENCODED_IOV_COMPRESSION_LZO ENCODED_IOV_COMPRESSION_LZO
+	ENCODED_IOV_COMPRESSION_ZSTD,
+#define ENCODED_IOV_COMPRESSION_ZSTD ENCODED_IOV_COMPRESSION_ZSTD
+	ENCODED_IOV_COMPRESSION_TYPES = ENCODED_IOV_COMPRESSION_ZSTD,
+};
+
+enum {
+	ENCODED_IOV_ENCRYPTION_NONE,
+#define ENCODED_IOV_ENCRYPTION_NONE ENCODED_IOV_ENCRYPTION_NONE
+	ENCODED_IOV_ENCRYPTION_TYPES = ENCODED_IOV_ENCRYPTION_NONE,
+};
+
+struct encoded_iov {
+	__aligned_u64 len;
+	__aligned_u64 unencoded_len;
+	__aligned_u64 unencoded_offset;
+	__u32 compression;
+	__u32 encryption;
+};
+
+#define ENCODED_IOV_SIZE_VER0 32
+
 /*
  * Flags for preadv2/pwritev2:
  */
@@ -299,8 +327,11 @@ typedef int __bitwise __kernel_rwf_t;
 /* per-IO O_APPEND */
 #define RWF_APPEND	((__force __kernel_rwf_t)0x00000010)
 
+/* encoded (e.g., compressed and/or encrypted) IO */
+#define RWF_ENCODED	((__force __kernel_rwf_t)0x00000020)
+
 /* mask of flags supported by the kernel */
 #define RWF_SUPPORTED	(RWF_HIPRI | RWF_DSYNC | RWF_SYNC | RWF_NOWAIT |\
-			 RWF_APPEND)
+			 RWF_APPEND | RWF_ENCODED)
 
 #endif /* _UAPI_LINUX_FS_H */
diff --git a/mm/filemap.c b/mm/filemap.c
index 1784478270e1..88a0bea618d2 100644
--- a/mm/filemap.c
+++ b/mm/filemap.c
@@ -2906,24 +2906,15 @@ static int generic_write_check_limits(struct file *file, loff_t pos,
 	return 0;
 }
 
-/*
- * Performs necessary checks before doing a write
- *
- * Can adjust writing position or amount of bytes to write.
- * Returns appropriate error code that caller should return or
- * zero in case that write should be allowed.
- */
-inline ssize_t generic_write_checks(struct kiocb *iocb, struct iov_iter *from)
+static int generic_write_checks_common(struct kiocb *iocb, loff_t *count)
 {
 	struct file *file = iocb->ki_filp;
 	struct inode *inode = file->f_mapping->host;
-	loff_t count;
-	int ret;
 
 	if (IS_SWAPFILE(inode))
 		return -ETXTBSY;
 
-	if (!iov_iter_count(from))
+	if (!*count)
 		return 0;
 
 	/* FIXME: this is for backwards compatibility with 2.4 */
@@ -2933,8 +2924,21 @@ inline ssize_t generic_write_checks(struct kiocb *iocb, struct iov_iter *from)
 	if ((iocb->ki_flags & IOCB_NOWAIT) && !(iocb->ki_flags & IOCB_DIRECT))
 		return -EINVAL;
 
-	count = iov_iter_count(from);
-	ret = generic_write_check_limits(file, iocb->ki_pos, &count);
+	return generic_write_check_limits(iocb->ki_filp, iocb->ki_pos, count);
+}
+
+/*
+ * Performs necessary checks before doing a write
+ *
+ * Can adjust writing position or amount of bytes to write.
+ * Returns a negative errno or the new number of bytes to write.
+ */
+inline ssize_t generic_write_checks(struct kiocb *iocb, struct iov_iter *from)
+{
+	loff_t count = iov_iter_count(from);
+	int ret;
+
+	ret = generic_write_checks_common(iocb, &count);
 	if (ret)
 		return ret;
 
@@ -2943,6 +2947,142 @@ inline ssize_t generic_write_checks(struct kiocb *iocb, struct iov_iter *from)
 }
 EXPORT_SYMBOL(generic_write_checks);
 
+/**
+ * generic_encoded_write_checks() - check an encoded write
+ * @iocb: I/O context.
+ * @encoded: Encoding metadata.
+ *
+ * This should be called by RWF_ENCODED write implementations rather than
+ * generic_write_checks(). Unlike generic_write_checks(), it returns -EFBIG
+ * instead of adjusting the size of the write.
+ *
+ * Return: 0 on success, -errno on error.
+ */
+int generic_encoded_write_checks(struct kiocb *iocb,
+				 const struct encoded_iov *encoded)
+{
+	loff_t count = encoded->len;
+	int ret;
+
+	if (!(iocb->ki_filp->f_flags & O_ALLOW_ENCODED))
+		return -EPERM;
+
+	ret = generic_write_checks_common(iocb, &count);
+	if (ret)
+		return ret;
+
+	if (count != encoded->len) {
+		/*
+		 * The write got truncated by generic_write_checks_common(). We
+		 * can't do a partial encoded write.
+		 */
+		return -EFBIG;
+	}
+	return 0;
+}
+EXPORT_SYMBOL(generic_encoded_write_checks);
+
+/**
+ * copy_encoded_iov_from_iter() - copy a &struct encoded_iov from userspace
+ * @encoded: Returned encoding metadata.
+ * @from: Source iterator.
+ *
+ * This copies in the &struct encoded_iov and does some basic sanity checks.
+ * This should always be used rather than a plain copy_from_iter(), as it does
+ * the proper handling for backward- and forward-compatibility.
+ *
+ * Return: 0 on success, -EFAULT if access to userspace failed, -E2BIG if the
+ *         copied structure contained non-zero fields that this kernel doesn't
+ *         support, -EINVAL if the copied structure was invalid.
+ */
+int copy_encoded_iov_from_iter(struct encoded_iov *encoded,
+			       struct iov_iter *from)
+{
+	size_t usize;
+	int ret;
+
+	usize = iov_iter_single_seg_count(from);
+	if (usize > PAGE_SIZE)
+		return -E2BIG;
+	if (usize < ENCODED_IOV_SIZE_VER0)
+		return -EINVAL;
+	ret = copy_struct_from_iter(encoded, sizeof(*encoded), from, usize);
+	if (ret)
+		return ret;
+
+	if (encoded->compression == ENCODED_IOV_COMPRESSION_NONE &&
+	    encoded->encryption == ENCODED_IOV_ENCRYPTION_NONE)
+		return -EINVAL;
+	if (encoded->compression > ENCODED_IOV_COMPRESSION_TYPES ||
+	    encoded->encryption > ENCODED_IOV_ENCRYPTION_TYPES)
+		return -EINVAL;
+	if (encoded->unencoded_offset > encoded->unencoded_len)
+		return -EINVAL;
+	if (encoded->len > encoded->unencoded_len - encoded->unencoded_offset)
+		return -EINVAL;
+	return 0;
+}
+EXPORT_SYMBOL(copy_encoded_iov_from_iter);
+
+/**
+ * generic_encoded_read_checks() - sanity check an RWF_ENCODED read
+ * @iocb: I/O context.
+ * @iter: Destination iterator for read.
+ *
+ * This should always be called by RWF_ENCODED read implementations before
+ * returning any data.
+ *
+ * Return: Number of bytes available to return encoded data in @iter on success,
+ *         -EPERM if the file was not opened with O_ALLOW_ENCODED, -EINVAL if
+ *         the size of the &struct encoded_iov iovec is invalid.
+ */
+ssize_t generic_encoded_read_checks(struct kiocb *iocb, struct iov_iter *iter)
+{
+	size_t usize;
+
+	if (!(iocb->ki_filp->f_flags & O_ALLOW_ENCODED))
+		return -EPERM;
+	usize = iov_iter_single_seg_count(iter);
+	if (usize > PAGE_SIZE || usize < ENCODED_IOV_SIZE_VER0)
+		return -EINVAL;
+	return iov_iter_count(iter) - usize;
+}
+EXPORT_SYMBOL(generic_encoded_read_checks);
+
+/**
+ * copy_encoded_iov_to_iter() - copy a &struct encoded_iov to userspace
+ * @encoded: Encoding metadata to return.
+ * @to: Destination iterator.
+ *
+ * This should always be used by RWF_ENCODED read implementations rather than a
+ * plain copy_to_iter(), as it does the proper handling for backward- and
+ * forward-compatibility. The iterator must be sanity-checked with
+ * generic_encoded_read_checks() before this is called.
+ *
+ * Return: 0 on success, -EFAULT if access to userspace failed, -E2BIG if there
+ *         were non-zero fields in @encoded that the user buffer could not
+ *         accommodate.
+ */
+int copy_encoded_iov_to_iter(const struct encoded_iov *encoded,
+			     struct iov_iter *to)
+{
+	size_t ksize = sizeof(*encoded);
+	size_t usize = iov_iter_single_seg_count(to);
+	size_t size = min(ksize, usize);
+
+	/* We already sanity-checked usize in generic_encoded_read_checks(). */
+
+	if (usize < ksize &&
+	    memchr_inv((char *)encoded + usize, 0, ksize - usize))
+		return -E2BIG;
+	if (copy_to_iter(encoded, size, to) != size ||
+	    (usize > ksize &&
+	     iov_iter_zero(usize - ksize, to) != usize - ksize))
+		return -EFAULT;
+	return 0;
+}
+EXPORT_SYMBOL(copy_encoded_iov_to_iter);
+
 /*
  * Performs necessary checks before doing a clone.
  *
-- 
2.25.1

Re: [PATCH v4 3/9] fs: add RWF_ENCODED for reading/writing compressed data

[Amir Goldstein]

On Sat, Feb 29, 2020 at 1:14 AM Omar Sandoval <osandov@osandov.com> wrote:
>
> From: Omar Sandoval <osandov@fb.com>
>
> Btrfs supports transparent compression: data written by the user can be
> compressed when written to disk and decompressed when read back.
> However, we'd like to add an interface to write pre-compressed data
> directly to the filesystem, and the matching interface to read
> compressed data without decompressing it. This adds support for
> so-called "encoded I/O" via preadv2() and pwritev2().
>
> A new RWF_ENCODED flags indicates that a read or write is "encoded". If
> this flag is set, iov[0].iov_base points to a struct encoded_iov which
> is used for metadata: namely, the compression algorithm, unencoded
> (i.e., decompressed) length, and what subrange of the unencoded data
> should be used (needed for truncated or hole-punched extents and when
> reading in the middle of an extent). For reads, the filesystem returns
> this information; for writes, the caller provides it to the filesystem.
> iov[0].iov_len must be set to sizeof(struct encoded_iov), which can be
> used to extend the interface in the future a la copy_struct_from_user().
> The remaining iovecs contain the encoded extent.
>
> This adds the VFS helpers for supporting encoded I/O and documentation
> for filesystem support.
>
> Signed-off-by: Omar Sandoval <osandov@fb.com>
> ---
>  Documentation/filesystems/encoded_io.rst |  74 ++++++++++
>  Documentation/filesystems/index.rst      |   1 +
>  include/linux/fs.h                       |  16 +++
>  include/uapi/linux/fs.h                  |  33 ++++-
>  mm/filemap.c                             | 166 +++++++++++++++++++++--
>  5 files changed, 276 insertions(+), 14 deletions(-)
>  create mode 100644 Documentation/filesystems/encoded_io.rst
>
> diff --git a/Documentation/filesystems/encoded_io.rst b/Documentation/filesystems/encoded_io.rst
> new file mode 100644
> index 000000000000..50405276d866
> --- /dev/null
> +++ b/Documentation/filesystems/encoded_io.rst
> @@ -0,0 +1,74 @@
> +===========
> +Encoded I/O
> +===========
> +
> +Encoded I/O is a mechanism for reading and writing encoded (e.g., compressed
> +and/or encrypted) data directly from/to the filesystem. The userspace interface
> +is thoroughly described in the :manpage:`encoded_io(7)` man page; this document
> +describes the requirements for filesystem support.
> +
> +First of all, a filesystem supporting encoded I/O must indicate this by setting
> +the ``FMODE_ENCODED_IO`` flag in its ``file_open`` file operation::
> +
> +    static int foo_file_open(struct inode *inode, struct file *filp)
> +    {
> +            ...
> +            filep->f_mode |= FMODE_ENCODED_IO;
> +            ...
> +    }
> +
> +Encoded I/O goes through ``read_iter`` and ``write_iter``, designated by the
> +``IOCB_ENCODED`` flag in ``kiocb->ki_flags``.
> +
> +Reads
> +=====
> +
> +Encoded ``read_iter`` should:
> +
> +1. Call ``generic_encoded_read_checks()`` to validate the file and buffers
> +   provided by userspace.
> +2. Initialize the ``encoded_iov`` appropriately.
> +3. Copy it to the user with ``copy_encoded_iov_to_iter()``.
> +4. Copy the encoded data to the user.
> +5. Advance ``kiocb->ki_pos`` by ``encoded_iov->len``.
> +6. Return the size of the encoded data read, not including the ``encoded_iov``.
> +
> +There are a few details to be aware of:
> +
> +* Encoded ``read_iter`` should support reading unencoded data if the extent is
> +  not encoded.
> +* If the buffers provided by the user are not large enough to contain an entire
> +  encoded extent, then ``read_iter`` should return ``-ENOBUFS``. This is to
> +  avoid confusing userspace with truncated data that cannot be properly
> +  decoded.
> +* Reads in the middle of an encoded extent can be returned by setting
> +  ``encoded_iov->unencoded_offset`` to non-zero.
> +* Truncated unencoded data (e.g., because the file does not end on a block
> +  boundary) may be returned by setting ``encoded_iov->len`` to a value smaller
> +  value than ``encoded_iov->unencoded_len - encoded_iov->unencoded_offset``.
> +
> +Writes
> +======
> +
> +Encoded ``write_iter`` should (in addition to the usual accounting/checks done
> +by ``write_iter``):
> +
> +1. Call ``copy_encoded_iov_from_iter()`` to get and validate the
> +   ``encoded_iov``.
> +2. Call ``generic_encoded_write_checks()`` instead of
> +   ``generic_write_checks()``.
> +3. Check that the provided encoding in ``encoded_iov`` is supported.
> +4. Advance ``kiocb->ki_pos`` by ``encoded_iov->len``.
> +5. Return the size of the encoded data written.
> +
> +Again, there are a few details:
> +
> +* Encoded ``write_iter`` doesn't need to support writing unencoded data.
> +* ``write_iter`` should either write all of the encoded data or none of it; it
> +  must not do partial writes.
> +* ``write_iter`` doesn't need to validate the encoded data; a subsequent read
> +  may return, e.g., ``-EIO`` if the data is not valid.
> +* The user may lie about the unencoded size of the data; a subsequent read
> +  should truncate or zero-extend the unencoded data rather than returning an
> +  error.
> +* Be careful of page cache coherency.
> diff --git a/Documentation/filesystems/index.rst b/Documentation/filesystems/index.rst
> index 386eaad008b2..e074a3f1f856 100644
> --- a/Documentation/filesystems/index.rst
> +++ b/Documentation/filesystems/index.rst
> @@ -37,6 +37,7 @@ filesystem implementations.
>     journalling
>     fscrypt
>     fsverity
> +   encoded_io
>
>  Filesystems
>  ===========
> diff --git a/include/linux/fs.h b/include/linux/fs.h
> index 3cd4fe6b845e..aa7efd3430d1 100644
> --- a/include/linux/fs.h
> +++ b/include/linux/fs.h
> @@ -175,6 +175,9 @@ typedef int (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
>  /* File does not contribute to nr_files count */
>  #define FMODE_NOACCOUNT                ((__force fmode_t)0x20000000)
>
> +/* File supports encoded IO */
> +#define FMODE_ENCODED_IO       ((__force fmode_t)0x40000000)
> +
>  /*
>   * Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector
>   * that indicates that they should check the contents of the iovec are
> @@ -314,6 +317,7 @@ enum rw_hint {
>  #define IOCB_SYNC              (1 << 5)
>  #define IOCB_WRITE             (1 << 6)
>  #define IOCB_NOWAIT            (1 << 7)
> +#define IOCB_ENCODED           (1 << 8)
>
>  struct kiocb {
>         struct file             *ki_filp;
> @@ -3109,6 +3113,13 @@ extern int sb_min_blocksize(struct super_block *, int);
>  extern int generic_file_mmap(struct file *, struct vm_area_struct *);
>  extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
>  extern ssize_t generic_write_checks(struct kiocb *, struct iov_iter *);
> +struct encoded_iov;
> +extern int generic_encoded_write_checks(struct kiocb *,
> +                                       const struct encoded_iov *);
> +extern int copy_encoded_iov_from_iter(struct encoded_iov *, struct iov_iter *);
> +extern ssize_t generic_encoded_read_checks(struct kiocb *, struct iov_iter *);
> +extern int copy_encoded_iov_to_iter(const struct encoded_iov *,
> +                                   struct iov_iter *);
>  extern int generic_remap_checks(struct file *file_in, loff_t pos_in,
>                                 struct file *file_out, loff_t pos_out,
>                                 loff_t *count, unsigned int remap_flags);
> @@ -3434,6 +3445,11 @@ static inline int kiocb_set_rw_flags(struct kiocb *ki, rwf_t flags)
>                         return -EOPNOTSUPP;
>                 ki->ki_flags |= IOCB_NOWAIT;
>         }
> +       if (flags & RWF_ENCODED) {
> +               if (!(ki->ki_filp->f_mode & FMODE_ENCODED_IO))
> +                       return -EOPNOTSUPP;
> +               ki->ki_flags |= IOCB_ENCODED;
> +       }
>         if (flags & RWF_HIPRI)
>                 ki->ki_flags |= IOCB_HIPRI;
>         if (flags & RWF_DSYNC)
> diff --git a/include/uapi/linux/fs.h b/include/uapi/linux/fs.h
> index 379a612f8f1d..f8c6c1e08def 100644
> --- a/include/uapi/linux/fs.h
> +++ b/include/uapi/linux/fs.h
> @@ -278,6 +278,34 @@ struct fsxattr {
>                                          SYNC_FILE_RANGE_WAIT_BEFORE | \
>                                          SYNC_FILE_RANGE_WAIT_AFTER)
>
> +enum {
> +       ENCODED_IOV_COMPRESSION_NONE,
> +#define ENCODED_IOV_COMPRESSION_NONE ENCODED_IOV_COMPRESSION_NONE
> +       ENCODED_IOV_COMPRESSION_ZLIB,
> +#define ENCODED_IOV_COMPRESSION_ZLIB ENCODED_IOV_COMPRESSION_ZLIB
> +       ENCODED_IOV_COMPRESSION_LZO,
> +#define ENCODED_IOV_COMPRESSION_LZO ENCODED_IOV_COMPRESSION_LZO
> +       ENCODED_IOV_COMPRESSION_ZSTD,
> +#define ENCODED_IOV_COMPRESSION_ZSTD ENCODED_IOV_COMPRESSION_ZSTD
> +       ENCODED_IOV_COMPRESSION_TYPES = ENCODED_IOV_COMPRESSION_ZSTD,
> +};
> +
> +enum {
> +       ENCODED_IOV_ENCRYPTION_NONE,
> +#define ENCODED_IOV_ENCRYPTION_NONE ENCODED_IOV_ENCRYPTION_NONE
> +       ENCODED_IOV_ENCRYPTION_TYPES = ENCODED_IOV_ENCRYPTION_NONE,
> +};
> +

What are those defines???

> +struct encoded_iov {
> +       __aligned_u64 len;
> +       __aligned_u64 unencoded_len;
> +       __aligned_u64 unencoded_offset;
> +       __u32 compression;
> +       __u32 encryption;
> +};
> +
[...]

> +/**
> + * copy_encoded_iov_from_iter() - copy a &struct encoded_iov from userspace
> + * @encoded: Returned encoding metadata.
> + * @from: Source iterator.
> + *
> + * This copies in the &struct encoded_iov and does some basic sanity checks.
> + * This should always be used rather than a plain copy_from_iter(), as it does
> + * the proper handling for backward- and forward-compatibility.
> + *
> + * Return: 0 on success, -EFAULT if access to userspace failed, -E2BIG if the
> + *         copied structure contained non-zero fields that this kernel doesn't
> + *         support, -EINVAL if the copied structure was invalid.
> + */
> +int copy_encoded_iov_from_iter(struct encoded_iov *encoded,
> +                              struct iov_iter *from)
> +{
> +       size_t usize;
> +       int ret;
> +
> +       usize = iov_iter_single_seg_count(from);
> +       if (usize > PAGE_SIZE)
> +               return -E2BIG;
> +       if (usize < ENCODED_IOV_SIZE_VER0)
> +               return -EINVAL;
> +       ret = copy_struct_from_iter(encoded, sizeof(*encoded), from, usize);
> +       if (ret)
> +               return ret;
> +
> +       if (encoded->compression == ENCODED_IOV_COMPRESSION_NONE &&
> +           encoded->encryption == ENCODED_IOV_ENCRYPTION_NONE)
> +               return -EINVAL;
> +       if (encoded->compression > ENCODED_IOV_COMPRESSION_TYPES ||
> +           encoded->encryption > ENCODED_IOV_ENCRYPTION_TYPES)
> +               return -EINVAL;
> +       if (encoded->unencoded_offset > encoded->unencoded_len)
> +               return -EINVAL;
> +       if (encoded->len > encoded->unencoded_len - encoded->unencoded_offset)
> +               return -EINVAL;
> +       return 0;
> +}

Repeating my comment from man page review:
It would be nice if a more granular error ENCODED_IOV_ERR_XXX code could be
set in the  encoded_iov struct.

Thanks,
Amir.

Re: [PATCH v4 3/9] fs: add RWF_ENCODED for reading/writing compressed data

[Omar Sandoval]

On Sat, Feb 29, 2020 at 12:40:58PM +0200, Amir Goldstein wrote:
> On Sat, Feb 29, 2020 at 1:14 AM Omar Sandoval <osandov@osandov.com> wrote:
> >
> > From: Omar Sandoval <osandov@fb.com>
> >
> > Btrfs supports transparent compression: data written by the user can be
> > compressed when written to disk and decompressed when read back.
> > However, we'd like to add an interface to write pre-compressed data
> > directly to the filesystem, and the matching interface to read
> > compressed data without decompressing it. This adds support for
> > so-called "encoded I/O" via preadv2() and pwritev2().
> >
> > A new RWF_ENCODED flags indicates that a read or write is "encoded". If
> > this flag is set, iov[0].iov_base points to a struct encoded_iov which
> > is used for metadata: namely, the compression algorithm, unencoded
> > (i.e., decompressed) length, and what subrange of the unencoded data
> > should be used (needed for truncated or hole-punched extents and when
> > reading in the middle of an extent). For reads, the filesystem returns
> > this information; for writes, the caller provides it to the filesystem.
> > iov[0].iov_len must be set to sizeof(struct encoded_iov), which can be
> > used to extend the interface in the future a la copy_struct_from_user().
> > The remaining iovecs contain the encoded extent.
> >
> > This adds the VFS helpers for supporting encoded I/O and documentation
> > for filesystem support.
> >
> > Signed-off-by: Omar Sandoval <osandov@fb.com>
> > ---
> >  Documentation/filesystems/encoded_io.rst |  74 ++++++++++
> >  Documentation/filesystems/index.rst      |   1 +
> >  include/linux/fs.h                       |  16 +++
> >  include/uapi/linux/fs.h                  |  33 ++++-
> >  mm/filemap.c                             | 166 +++++++++++++++++++++--
> >  5 files changed, 276 insertions(+), 14 deletions(-)
> >  create mode 100644 Documentation/filesystems/encoded_io.rst
> >
> > diff --git a/Documentation/filesystems/encoded_io.rst b/Documentation/filesystems/encoded_io.rst
> > new file mode 100644
> > index 000000000000..50405276d866
> > --- /dev/null
> > +++ b/Documentation/filesystems/encoded_io.rst
> > @@ -0,0 +1,74 @@
> > +===========
> > +Encoded I/O
> > +===========
> > +
> > +Encoded I/O is a mechanism for reading and writing encoded (e.g., compressed
> > +and/or encrypted) data directly from/to the filesystem. The userspace interface
> > +is thoroughly described in the :manpage:`encoded_io(7)` man page; this document
> > +describes the requirements for filesystem support.
> > +
> > +First of all, a filesystem supporting encoded I/O must indicate this by setting
> > +the ``FMODE_ENCODED_IO`` flag in its ``file_open`` file operation::
> > +
> > +    static int foo_file_open(struct inode *inode, struct file *filp)
> > +    {
> > +            ...
> > +            filep->f_mode |= FMODE_ENCODED_IO;
> > +            ...
> > +    }
> > +
> > +Encoded I/O goes through ``read_iter`` and ``write_iter``, designated by the
> > +``IOCB_ENCODED`` flag in ``kiocb->ki_flags``.
> > +
> > +Reads
> > +=====
> > +
> > +Encoded ``read_iter`` should:
> > +
> > +1. Call ``generic_encoded_read_checks()`` to validate the file and buffers
> > +   provided by userspace.
> > +2. Initialize the ``encoded_iov`` appropriately.
> > +3. Copy it to the user with ``copy_encoded_iov_to_iter()``.
> > +4. Copy the encoded data to the user.
> > +5. Advance ``kiocb->ki_pos`` by ``encoded_iov->len``.
> > +6. Return the size of the encoded data read, not including the ``encoded_iov``.
> > +
> > +There are a few details to be aware of:
> > +
> > +* Encoded ``read_iter`` should support reading unencoded data if the extent is
> > +  not encoded.
> > +* If the buffers provided by the user are not large enough to contain an entire
> > +  encoded extent, then ``read_iter`` should return ``-ENOBUFS``. This is to
> > +  avoid confusing userspace with truncated data that cannot be properly
> > +  decoded.
> > +* Reads in the middle of an encoded extent can be returned by setting
> > +  ``encoded_iov->unencoded_offset`` to non-zero.
> > +* Truncated unencoded data (e.g., because the file does not end on a block
> > +  boundary) may be returned by setting ``encoded_iov->len`` to a value smaller
> > +  value than ``encoded_iov->unencoded_len - encoded_iov->unencoded_offset``.
> > +
> > +Writes
> > +======
> > +
> > +Encoded ``write_iter`` should (in addition to the usual accounting/checks done
> > +by ``write_iter``):
> > +
> > +1. Call ``copy_encoded_iov_from_iter()`` to get and validate the
> > +   ``encoded_iov``.
> > +2. Call ``generic_encoded_write_checks()`` instead of
> > +   ``generic_write_checks()``.
> > +3. Check that the provided encoding in ``encoded_iov`` is supported.
> > +4. Advance ``kiocb->ki_pos`` by ``encoded_iov->len``.
> > +5. Return the size of the encoded data written.
> > +
> > +Again, there are a few details:
> > +
> > +* Encoded ``write_iter`` doesn't need to support writing unencoded data.
> > +* ``write_iter`` should either write all of the encoded data or none of it; it
> > +  must not do partial writes.
> > +* ``write_iter`` doesn't need to validate the encoded data; a subsequent read
> > +  may return, e.g., ``-EIO`` if the data is not valid.
> > +* The user may lie about the unencoded size of the data; a subsequent read
> > +  should truncate or zero-extend the unencoded data rather than returning an
> > +  error.
> > +* Be careful of page cache coherency.
> > diff --git a/Documentation/filesystems/index.rst b/Documentation/filesystems/index.rst
> > index 386eaad008b2..e074a3f1f856 100644
> > --- a/Documentation/filesystems/index.rst
> > +++ b/Documentation/filesystems/index.rst
> > @@ -37,6 +37,7 @@ filesystem implementations.
> >     journalling
> >     fscrypt
> >     fsverity
> > +   encoded_io
> >
> >  Filesystems
> >  ===========
> > diff --git a/include/linux/fs.h b/include/linux/fs.h
> > index 3cd4fe6b845e..aa7efd3430d1 100644
> > --- a/include/linux/fs.h
> > +++ b/include/linux/fs.h
> > @@ -175,6 +175,9 @@ typedef int (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
> >  /* File does not contribute to nr_files count */
> >  #define FMODE_NOACCOUNT                ((__force fmode_t)0x20000000)
> >
> > +/* File supports encoded IO */
> > +#define FMODE_ENCODED_IO       ((__force fmode_t)0x40000000)
> > +
> >  /*
> >   * Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector
> >   * that indicates that they should check the contents of the iovec are
> > @@ -314,6 +317,7 @@ enum rw_hint {
> >  #define IOCB_SYNC              (1 << 5)
> >  #define IOCB_WRITE             (1 << 6)
> >  #define IOCB_NOWAIT            (1 << 7)
> > +#define IOCB_ENCODED           (1 << 8)
> >
> >  struct kiocb {
> >         struct file             *ki_filp;
> > @@ -3109,6 +3113,13 @@ extern int sb_min_blocksize(struct super_block *, int);
> >  extern int generic_file_mmap(struct file *, struct vm_area_struct *);
> >  extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
> >  extern ssize_t generic_write_checks(struct kiocb *, struct iov_iter *);
> > +struct encoded_iov;
> > +extern int generic_encoded_write_checks(struct kiocb *,
> > +                                       const struct encoded_iov *);
> > +extern int copy_encoded_iov_from_iter(struct encoded_iov *, struct iov_iter *);
> > +extern ssize_t generic_encoded_read_checks(struct kiocb *, struct iov_iter *);
> > +extern int copy_encoded_iov_to_iter(const struct encoded_iov *,
> > +                                   struct iov_iter *);
> >  extern int generic_remap_checks(struct file *file_in, loff_t pos_in,
> >                                 struct file *file_out, loff_t pos_out,
> >                                 loff_t *count, unsigned int remap_flags);
> > @@ -3434,6 +3445,11 @@ static inline int kiocb_set_rw_flags(struct kiocb *ki, rwf_t flags)
> >                         return -EOPNOTSUPP;
> >                 ki->ki_flags |= IOCB_NOWAIT;
> >         }
> > +       if (flags & RWF_ENCODED) {
> > +               if (!(ki->ki_filp->f_mode & FMODE_ENCODED_IO))
> > +                       return -EOPNOTSUPP;
> > +               ki->ki_flags |= IOCB_ENCODED;
> > +       }
> >         if (flags & RWF_HIPRI)
> >                 ki->ki_flags |= IOCB_HIPRI;
> >         if (flags & RWF_DSYNC)
> > diff --git a/include/uapi/linux/fs.h b/include/uapi/linux/fs.h
> > index 379a612f8f1d..f8c6c1e08def 100644
> > --- a/include/uapi/linux/fs.h
> > +++ b/include/uapi/linux/fs.h
> > @@ -278,6 +278,34 @@ struct fsxattr {
> >                                          SYNC_FILE_RANGE_WAIT_BEFORE | \
> >                                          SYNC_FILE_RANGE_WAIT_AFTER)
> >
> > +enum {
> > +       ENCODED_IOV_COMPRESSION_NONE,
> > +#define ENCODED_IOV_COMPRESSION_NONE ENCODED_IOV_COMPRESSION_NONE
> > +       ENCODED_IOV_COMPRESSION_ZLIB,
> > +#define ENCODED_IOV_COMPRESSION_ZLIB ENCODED_IOV_COMPRESSION_ZLIB
> > +       ENCODED_IOV_COMPRESSION_LZO,
> > +#define ENCODED_IOV_COMPRESSION_LZO ENCODED_IOV_COMPRESSION_LZO
> > +       ENCODED_IOV_COMPRESSION_ZSTD,
> > +#define ENCODED_IOV_COMPRESSION_ZSTD ENCODED_IOV_COMPRESSION_ZSTD
> > +       ENCODED_IOV_COMPRESSION_TYPES = ENCODED_IOV_COMPRESSION_ZSTD,
> > +};
> > +
> > +enum {
> > +       ENCODED_IOV_ENCRYPTION_NONE,
> > +#define ENCODED_IOV_ENCRYPTION_NONE ENCODED_IOV_ENCRYPTION_NONE
> > +       ENCODED_IOV_ENCRYPTION_TYPES = ENCODED_IOV_ENCRYPTION_NONE,
> > +};
> > +
> 
> What are those defines???

They're so you can check whether an enum value is defined in the UAPI
headers via ifdef. E.g., if we were to add
ENCODED_IOV_COMPRESSION_SOME_NEW_ALGORITHM, applications could use:

#ifndef ENCODED_IOV_COMPRESSION_SOME_NEW_ALGORITHM
#define ENCODED_IOV_COMPRESSION_SOME_NEW_ALGORITHM 4
#endif

In my experience, this makes dealing with lagging UAPI headers less
annoying. This is done elsewhere in UAPI headers (e.g.,
include/uapi/linux/in.h).

[PATCH v4 4/9] btrfs: don't advance offset for compressed bios in btrfs_csum_one_bio()

[Omar Sandoval]

From: Omar Sandoval <osandov@fb.com>

btrfs_csum_one_bio() loops over each filesystem block in the bio while
keeping a cursor of its current logical position in the file in order to
look up the ordered extent to add the checksums to. However, this
doesn't make much sense for compressed extents, as a sector on disk does
not correspond to a sector of decompressed file data. It happens to work
because 1) the compressed bio always covers one ordered extent and 2)
the size of the bio is always less than the size of the ordered extent.
However, the second point will not always be true for encoded writes.

Let's add a boolean parameter to btrfs_csum_one_bio() to indicate that
it can assume that the bio only covers one ordered extent. Since we're
already changing the signature, let's get rid of the contig parameter
and make it implied by the offset parameter, similar to the change we
recently made to btrfs_lookup_bio_sums(). Additionally, let's rename
nr_sectors to blockcount to make it clear that it's the number of
filesystem blocks, not the number of 512-byte sectors.

Signed-off-by: Omar Sandoval <osandov@fb.com>
---
 fs/btrfs/compression.c |  5 +++--
 fs/btrfs/ctree.h       |  2 +-
 fs/btrfs/file-item.c   | 35 ++++++++++++++++-------------------
 fs/btrfs/inode.c       |  8 ++++----
 4 files changed, 24 insertions(+), 26 deletions(-)

diff --git a/fs/btrfs/compression.c b/fs/btrfs/compression.c
index 9ab610cc9114..b66846272971 100644
--- a/fs/btrfs/compression.c
+++ b/fs/btrfs/compression.c
@@ -477,7 +477,8 @@ blk_status_t btrfs_submit_compressed_write(struct inode *inode, u64 start,
 			BUG_ON(ret); /* -ENOMEM */
 
 			if (!skip_sum) {
-				ret = btrfs_csum_one_bio(inode, bio, start, 1);
+				ret = btrfs_csum_one_bio(inode, bio, start,
+							 true);
 				BUG_ON(ret); /* -ENOMEM */
 			}
 
@@ -509,7 +510,7 @@ blk_status_t btrfs_submit_compressed_write(struct inode *inode, u64 start,
 	BUG_ON(ret); /* -ENOMEM */
 
 	if (!skip_sum) {
-		ret = btrfs_csum_one_bio(inode, bio, start, 1);
+		ret = btrfs_csum_one_bio(inode, bio, start, true);
 		BUG_ON(ret); /* -ENOMEM */
 	}
 
diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h
index 36df977b64d9..d97983ab94fe 100644
--- a/fs/btrfs/ctree.h
+++ b/fs/btrfs/ctree.h
@@ -2851,7 +2851,7 @@ int btrfs_csum_file_blocks(struct btrfs_trans_handle *trans,
 			   struct btrfs_root *root,
 			   struct btrfs_ordered_sum *sums);
 blk_status_t btrfs_csum_one_bio(struct inode *inode, struct bio *bio,
-		       u64 file_start, int contig);
+				u64 offset, bool one_ordered);
 int btrfs_lookup_csums_range(struct btrfs_root *root, u64 start, u64 end,
 			     struct list_head *list, int search_commit);
 void btrfs_extent_item_to_extent_map(struct btrfs_inode *inode,
diff --git a/fs/btrfs/file-item.c b/fs/btrfs/file-item.c
index c2f365662d55..fe4f0700e033 100644
--- a/fs/btrfs/file-item.c
+++ b/fs/btrfs/file-item.c
@@ -422,28 +422,28 @@ int btrfs_lookup_csums_range(struct btrfs_root *root, u64 start, u64 end,
  * btrfs_csum_one_bio - Calculates checksums of the data contained inside a bio
  * @inode:	 Owner of the data inside the bio
  * @bio:	 Contains the data to be checksummed
- * @file_start:  offset in file this bio begins to describe
- * @contig:	 Boolean. If true/1 means all bio vecs in this bio are
- *		 contiguous and they begin at @file_start in the file. False/0
- *		 means this bio can contains potentially discontigous bio vecs
- *		 so the logical offset of each should be calculated separately.
+ * @offset:      If (u64)-1, @bio may contain discontiguous bio vecs, so the
+ *               file offsets are determined from the page offsets in the bio.
+ *               Otherwise, this is the starting file offset of the bio vecs in
+ *               @bio, which must be contiguous.
+ * @one_ordered: If true, @bio only refers to one ordered extent.
  */
 blk_status_t btrfs_csum_one_bio(struct inode *inode, struct bio *bio,
-		       u64 file_start, int contig)
+				u64 offset, bool one_ordered)
 {
 	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
 	SHASH_DESC_ON_STACK(shash, fs_info->csum_shash);
 	struct btrfs_ordered_sum *sums;
 	struct btrfs_ordered_extent *ordered = NULL;
+	const bool page_offsets = (offset == (u64)-1);
 	char *data;
 	struct bvec_iter iter;
 	struct bio_vec bvec;
 	int index;
-	int nr_sectors;
+	int blockcount;
 	unsigned long total_bytes = 0;
 	unsigned long this_sum_bytes = 0;
 	int i;
-	u64 offset;
 	unsigned nofs_flag;
 	const u16 csum_size = btrfs_super_csum_size(fs_info->super_copy);
 
@@ -458,18 +458,13 @@ blk_status_t btrfs_csum_one_bio(struct inode *inode, struct bio *bio,
 	sums->len = bio->bi_iter.bi_size;
 	INIT_LIST_HEAD(&sums->list);
 
-	if (contig)
-		offset = file_start;
-	else
-		offset = 0; /* shut up gcc */
-
 	sums->bytenr = (u64)bio->bi_iter.bi_sector << 9;
 	index = 0;
 
 	shash->tfm = fs_info->csum_shash;
 
 	bio_for_each_segment(bvec, bio, iter) {
-		if (!contig)
+		if (page_offsets)
 			offset = page_offset(bvec.bv_page) + bvec.bv_offset;
 
 		if (!ordered) {
@@ -477,13 +472,14 @@ blk_status_t btrfs_csum_one_bio(struct inode *inode, struct bio *bio,
 			BUG_ON(!ordered); /* Logic error */
 		}
 
-		nr_sectors = BTRFS_BYTES_TO_BLKS(fs_info,
+		blockcount = BTRFS_BYTES_TO_BLKS(fs_info,
 						 bvec.bv_len + fs_info->sectorsize
 						 - 1);
 
-		for (i = 0; i < nr_sectors; i++) {
-			if (offset >= ordered->file_offset + ordered->num_bytes ||
-			    offset < ordered->file_offset) {
+		for (i = 0; i < blockcount; i++) {
+			if (!one_ordered &&
+			    (offset >= ordered->file_offset + ordered->num_bytes ||
+			     offset < ordered->file_offset)) {
 				unsigned long bytes_left;
 
 				sums->len = this_sum_bytes;
@@ -515,7 +511,8 @@ blk_status_t btrfs_csum_one_bio(struct inode *inode, struct bio *bio,
 			kunmap_atomic(data);
 			crypto_shash_final(shash, (char *)(sums->sums + index));
 			index += csum_size;
-			offset += fs_info->sectorsize;
+			if (!one_ordered)
+				offset += fs_info->sectorsize;
 			this_sum_bytes += fs_info->sectorsize;
 			total_bytes += fs_info->sectorsize;
 		}
diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
index 1ccb3f8d528d..303613e6ec38 100644
--- a/fs/btrfs/inode.c
+++ b/fs/btrfs/inode.c
@@ -2075,7 +2075,7 @@ static blk_status_t btrfs_submit_bio_start(void *private_data, struct bio *bio,
 	struct inode *inode = private_data;
 	blk_status_t ret = 0;
 
-	ret = btrfs_csum_one_bio(inode, bio, 0, 0);
+	ret = btrfs_csum_one_bio(inode, bio, (u64)-1, false);
 	BUG_ON(ret); /* -ENOMEM */
 	return 0;
 }
@@ -2140,7 +2140,7 @@ static blk_status_t btrfs_submit_bio_hook(struct inode *inode, struct bio *bio,
 					  0, inode, btrfs_submit_bio_start);
 		goto out;
 	} else if (!skip_sum) {
-		ret = btrfs_csum_one_bio(inode, bio, 0, 0);
+		ret = btrfs_csum_one_bio(inode, bio, (u64)-1, false);
 		if (ret)
 			goto out;
 	}
@@ -7731,7 +7731,7 @@ static blk_status_t btrfs_submit_bio_start_direct_io(void *private_data,
 {
 	struct inode *inode = private_data;
 	blk_status_t ret;
-	ret = btrfs_csum_one_bio(inode, bio, offset, 1);
+	ret = btrfs_csum_one_bio(inode, bio, offset, false);
 	BUG_ON(ret); /* -ENOMEM */
 	return 0;
 }
@@ -7838,7 +7838,7 @@ static inline blk_status_t btrfs_submit_dio_bio(struct bio *bio,
 		 * If we aren't doing async submit, calculate the csum of the
 		 * bio now.
 		 */
-		ret = btrfs_csum_one_bio(inode, bio, file_offset, 1);
+		ret = btrfs_csum_one_bio(inode, bio, file_offset, false);
 		if (ret)
 			goto err;
 	} else {
-- 
2.25.1

[PATCH v4 5/9] btrfs: add ram_bytes and offset to btrfs_ordered_extent

[Omar Sandoval]

From: Omar Sandoval <osandov@fb.com>

Currently, we only create ordered extents when ram_bytes == num_bytes
and offset == 0. However, RWF_ENCODED writes may create extents which
only refer to a subset of the full unencoded extent, so we need to plumb
these fields through the ordered extent infrastructure and pass them
down to insert_reserved_file_extent().

Since we're changing the btrfs_add_ordered_extent* signature, let's get
rid of the trivial wrappers and add a kernel-doc.

Reviewed-by: Nikolay Borisov <nborisov@suse.com>
Signed-off-by: Omar Sandoval <osandov@fb.com>
---
 fs/btrfs/inode.c        | 65 +++++++++++++++++++++++------------------
 fs/btrfs/ordered-data.c | 65 +++++++++++++++--------------------------
 fs/btrfs/ordered-data.h | 16 ++++------
 3 files changed, 67 insertions(+), 79 deletions(-)

diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
index 303613e6ec38..0d5b4e14f815 100644
--- a/fs/btrfs/inode.c
+++ b/fs/btrfs/inode.c
@@ -848,13 +848,12 @@ static noinline void submit_compressed_extents(struct async_chunk *async_chunk)
 			goto out_free_reserve;
 		free_extent_map(em);
 
-		ret = btrfs_add_ordered_extent_compress(inode,
-						async_extent->start,
-						ins.objectid,
-						async_extent->ram_size,
-						ins.offset,
-						BTRFS_ORDERED_COMPRESSED,
-						async_extent->compress_type);
+		ret = btrfs_add_ordered_extent(inode, async_extent->start,
+					       async_extent->ram_size,
+					       async_extent->ram_size,
+					       ins.objectid, ins.offset, 0,
+					       1 << BTRFS_ORDERED_COMPRESSED,
+					       async_extent->compress_type);
 		if (ret) {
 			btrfs_drop_extent_cache(BTRFS_I(inode),
 						async_extent->start,
@@ -1049,8 +1048,9 @@ static noinline int cow_file_range(struct inode *inode,
 		}
 		free_extent_map(em);
 
-		ret = btrfs_add_ordered_extent(inode, start, ins.objectid,
-					       ram_size, cur_alloc_size, 0);
+		ret = btrfs_add_ordered_extent(inode, start, ram_size, ram_size,
+					       ins.objectid, cur_alloc_size, 0,
+					       0, BTRFS_COMPRESS_NONE);
 		if (ret)
 			goto out_drop_extent_cache;
 
@@ -1623,10 +1623,11 @@ static noinline int run_delalloc_nocow(struct inode *inode,
 				goto error;
 			}
 			free_extent_map(em);
-			ret = btrfs_add_ordered_extent(inode, cur_offset,
-						       disk_bytenr, num_bytes,
-						       num_bytes,
-						       BTRFS_ORDERED_PREALLOC);
+			ret = btrfs_add_ordered_extent(inode,
+					cur_offset, num_bytes, num_bytes,
+					disk_bytenr, num_bytes, 0,
+					1 << BTRFS_ORDERED_PREALLOC,
+					BTRFS_COMPRESS_NONE);
 			if (ret) {
 				btrfs_drop_extent_cache(BTRFS_I(inode),
 							cur_offset,
@@ -1636,9 +1637,11 @@ static noinline int run_delalloc_nocow(struct inode *inode,
 			}
 		} else {
 			ret = btrfs_add_ordered_extent(inode, cur_offset,
+						       num_bytes, num_bytes,
 						       disk_bytenr, num_bytes,
-						       num_bytes,
-						       BTRFS_ORDERED_NOCOW);
+						       0,
+						       1 << BTRFS_ORDERED_NOCOW,
+						       BTRFS_COMPRESS_NONE);
 			if (ret)
 				goto error;
 		}
@@ -2380,7 +2383,7 @@ int btrfs_writepage_cow_fixup(struct page *page, u64 start, u64 end)
 static int insert_reserved_file_extent(struct btrfs_trans_handle *trans,
 				       struct inode *inode, u64 file_pos,
 				       u64 disk_bytenr, u64 disk_num_bytes,
-				       u64 num_bytes, u64 ram_bytes,
+				       u64 offset, u64 num_bytes, u64 ram_bytes,
 				       u8 compression, u8 encryption,
 				       u16 other_encoding, int extent_type)
 {
@@ -2430,7 +2433,7 @@ static int insert_reserved_file_extent(struct btrfs_trans_handle *trans,
 	btrfs_set_file_extent_type(leaf, fi, extent_type);
 	btrfs_set_file_extent_disk_bytenr(leaf, fi, disk_bytenr);
 	btrfs_set_file_extent_disk_num_bytes(leaf, fi, disk_num_bytes);
-	btrfs_set_file_extent_offset(leaf, fi, 0);
+	btrfs_set_file_extent_offset(leaf, fi, offset);
 	btrfs_set_file_extent_num_bytes(leaf, fi, num_bytes);
 	btrfs_set_file_extent_ram_bytes(leaf, fi, ram_bytes);
 	btrfs_set_file_extent_compression(leaf, fi, compression);
@@ -2456,7 +2459,8 @@ static int insert_reserved_file_extent(struct btrfs_trans_handle *trans,
 	qg_released = ret;
 	ret = btrfs_alloc_reserved_file_extent(trans, root,
 					       btrfs_ino(BTRFS_I(inode)),
-					       file_pos, qg_released, &ins);
+					       file_pos - offset, qg_released,
+					       &ins);
 out:
 	btrfs_free_path(path);
 
@@ -2493,7 +2497,8 @@ static int btrfs_finish_ordered_io(struct btrfs_ordered_extent *ordered_extent)
 	u64 start, end;
 	int compress_type = 0;
 	int ret = 0;
-	u64 logical_len = ordered_extent->num_bytes;
+	u64 num_bytes = ordered_extent->num_bytes;
+	u64 ram_bytes = ordered_extent->ram_bytes;
 	bool freespace_inode;
 	bool truncated = false;
 	bool range_locked = false;
@@ -2520,9 +2525,9 @@ static int btrfs_finish_ordered_io(struct btrfs_ordered_extent *ordered_extent)
 
 	if (test_bit(BTRFS_ORDERED_TRUNCATED, &ordered_extent->flags)) {
 		truncated = true;
-		logical_len = ordered_extent->truncated_len;
+		num_bytes = ram_bytes = ordered_extent->truncated_len;
 		/* Truncated the entire extent, don't bother adding */
-		if (!logical_len)
+		if (!num_bytes)
 			goto out;
 	}
 
@@ -2577,13 +2582,14 @@ static int btrfs_finish_ordered_io(struct btrfs_ordered_extent *ordered_extent)
 		ret = btrfs_mark_extent_written(trans, BTRFS_I(inode),
 						ordered_extent->file_offset,
 						ordered_extent->file_offset +
-						logical_len);
+						num_bytes);
 	} else {
 		BUG_ON(root == fs_info->tree_root);
 		ret = insert_reserved_file_extent(trans, inode, start,
 						ordered_extent->disk_bytenr,
 						ordered_extent->disk_num_bytes,
-						logical_len, logical_len,
+						ordered_extent->offset,
+						num_bytes, ram_bytes,
 						compress_type, 0, 0,
 						BTRFS_FILE_EXTENT_REG);
 		if (!ret) {
@@ -2631,7 +2637,7 @@ static int btrfs_finish_ordered_io(struct btrfs_ordered_extent *ordered_extent)
 		u64 unwritten_start = start;
 
 		if (truncated)
-			unwritten_start += logical_len;
+			unwritten_start += num_bytes;
 		clear_extent_uptodate(io_tree, unwritten_start, end, NULL);
 
 		/* Drop the cache for the part of the extent we didn't write. */
@@ -2647,7 +2653,7 @@ static int btrfs_finish_ordered_io(struct btrfs_ordered_extent *ordered_extent)
 		 * errored out then we don't need to do this as the accounting
 		 * has already been done.
 		 */
-		if ((ret || !logical_len) &&
+		if ((ret || !num_bytes) &&
 		    clear_reserved_extent &&
 		    !test_bit(BTRFS_ORDERED_NOCOW, &ordered_extent->flags) &&
 		    !test_bit(BTRFS_ORDERED_PREALLOC, &ordered_extent->flags)) {
@@ -6752,8 +6758,11 @@ static struct extent_map *btrfs_create_dio_extent(struct inode *inode,
 		if (IS_ERR(em))
 			goto out;
 	}
-	ret = btrfs_add_ordered_extent_dio(inode, start, block_start,
-					   len, block_len, type);
+	ret = btrfs_add_ordered_extent(inode, start, len, len, block_start,
+				       block_len, 0,
+				       (1 << type) |
+				       (1 << BTRFS_ORDERED_DIRECT),
+				       BTRFS_COMPRESS_NONE);
 	if (ret) {
 		if (em) {
 			free_extent_map(em);
@@ -9873,7 +9882,7 @@ static int __btrfs_prealloc_file_range(struct inode *inode, int mode,
 		last_alloc = ins.offset;
 		ret = insert_reserved_file_extent(trans, inode,
 						  cur_offset, ins.objectid,
-						  ins.offset, ins.offset,
+						  ins.offset, 0, ins.offset,
 						  ins.offset, 0, 0, 0,
 						  BTRFS_FILE_EXTENT_PREALLOC);
 		if (ret) {
diff --git a/fs/btrfs/ordered-data.c b/fs/btrfs/ordered-data.c
index a65f189a5b94..9a5f35d35fa9 100644
--- a/fs/btrfs/ordered-data.c
+++ b/fs/btrfs/ordered-data.c
@@ -152,15 +152,27 @@ static inline struct rb_node *tree_search(struct btrfs_ordered_inode_tree *tree,
 	return ret;
 }
 
-/* allocate and add a new ordered_extent into the per-inode tree.
+/**
+ * btrfs_add_ordered_extent - Add an ordered extent to the per-inode tree.
+ * @inode: inode that this extent is for.
+ * @file_offset: Logical offset in file where the extent starts.
+ * @num_bytes: Logical length of extent in file.
+ * @ram_bytes: Full length of unencoded data.
+ * @disk_bytenr: Offset of extent on disk.
+ * @disk_num_bytes: Size of extent on disk.
+ * @offset: Offset into unencoded data where file data starts.
+ * @flags: Flags specifying type of extent (1 << BTRFS_ORDERED_*).
+ * @compress_type: Compression algorithm used for data.
  *
- * The tree is given a single reference on the ordered extent that was
- * inserted.
+ * Most of these parameters correspond to &struct btrfs_file_extent_item. The
+ * tree is given a single reference on the ordered extent that was inserted.
+ *
+ * Return: 0 or -ENOMEM.
  */
-static int __btrfs_add_ordered_extent(struct inode *inode, u64 file_offset,
-				      u64 disk_bytenr, u64 num_bytes,
-				      u64 disk_num_bytes, int type, int dio,
-				      int compress_type)
+int btrfs_add_ordered_extent(struct inode *inode, u64 file_offset,
+			     u64 num_bytes, u64 ram_bytes, u64 disk_bytenr,
+			     u64 disk_num_bytes, u64 offset, int flags,
+			     int compress_type)
 {
 	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
 	struct btrfs_root *root = BTRFS_I(inode)->root;
@@ -174,20 +186,19 @@ static int __btrfs_add_ordered_extent(struct inode *inode, u64 file_offset,
 		return -ENOMEM;
 
 	entry->file_offset = file_offset;
-	entry->disk_bytenr = disk_bytenr;
 	entry->num_bytes = num_bytes;
+	entry->ram_bytes = ram_bytes;
+	entry->disk_bytenr = disk_bytenr;
 	entry->disk_num_bytes = disk_num_bytes;
+	entry->offset = offset;
 	entry->bytes_left = num_bytes;
 	entry->inode = igrab(inode);
 	entry->compress_type = compress_type;
 	entry->truncated_len = (u64)-1;
-	if (type != BTRFS_ORDERED_IO_DONE && type != BTRFS_ORDERED_COMPLETE)
-		set_bit(type, &entry->flags);
-
-	if (dio) {
+	entry->flags = flags;
+	if (flags & (1 << BTRFS_ORDERED_DIRECT)) {
 		percpu_counter_add_batch(&fs_info->dio_bytes, num_bytes,
 					 fs_info->delalloc_batch);
-		set_bit(BTRFS_ORDERED_DIRECT, &entry->flags);
 	}
 
 	/* one ref for the tree */
@@ -235,34 +246,6 @@ static int __btrfs_add_ordered_extent(struct inode *inode, u64 file_offset,
 	return 0;
 }
 
-int btrfs_add_ordered_extent(struct inode *inode, u64 file_offset,
-			     u64 disk_bytenr, u64 num_bytes, u64 disk_num_bytes,
-			     int type)
-{
-	return __btrfs_add_ordered_extent(inode, file_offset, disk_bytenr,
-					  num_bytes, disk_num_bytes, type, 0,
-					  BTRFS_COMPRESS_NONE);
-}
-
-int btrfs_add_ordered_extent_dio(struct inode *inode, u64 file_offset,
-				 u64 disk_bytenr, u64 num_bytes,
-				 u64 disk_num_bytes, int type)
-{
-	return __btrfs_add_ordered_extent(inode, file_offset, disk_bytenr,
-					  num_bytes, disk_num_bytes, type, 1,
-					  BTRFS_COMPRESS_NONE);
-}
-
-int btrfs_add_ordered_extent_compress(struct inode *inode, u64 file_offset,
-				      u64 disk_bytenr, u64 num_bytes,
-				      u64 disk_num_bytes, int type,
-				      int compress_type)
-{
-	return __btrfs_add_ordered_extent(inode, file_offset, disk_bytenr,
-					  num_bytes, disk_num_bytes, type, 0,
-					  compress_type);
-}
-
 /*
  * Add a struct btrfs_ordered_sum into the list of checksums to be inserted
  * when an ordered extent is finished.  If the list covers more than one
diff --git a/fs/btrfs/ordered-data.h b/fs/btrfs/ordered-data.h
index 3beb4da4ab41..ef528fef5841 100644
--- a/fs/btrfs/ordered-data.h
+++ b/fs/btrfs/ordered-data.h
@@ -71,9 +71,11 @@ struct btrfs_ordered_extent {
 	 * These fields directly correspond to the same fields in
 	 * btrfs_file_extent_item.
 	 */
-	u64 disk_bytenr;
 	u64 num_bytes;
+	u64 ram_bytes;
+	u64 disk_bytenr;
 	u64 disk_num_bytes;
+	u64 offset;
 
 	/* number of bytes that still need writing */
 	u64 bytes_left;
@@ -160,15 +162,9 @@ int btrfs_dec_test_first_ordered_pending(struct inode *inode,
 				   u64 *file_offset, u64 io_size,
 				   int uptodate);
 int btrfs_add_ordered_extent(struct inode *inode, u64 file_offset,
-			     u64 disk_bytenr, u64 num_bytes, u64 disk_num_bytes,
-			     int type);
-int btrfs_add_ordered_extent_dio(struct inode *inode, u64 file_offset,
-				 u64 disk_bytenr, u64 num_bytes,
-				 u64 disk_num_bytes, int type);
-int btrfs_add_ordered_extent_compress(struct inode *inode, u64 file_offset,
-				      u64 disk_bytenr, u64 num_bytes,
-				      u64 disk_num_bytes, int type,
-				      int compress_type);
+			     u64 num_bytes, u64 ram_bytes, u64 disk_bytenr,
+			     u64 disk_num_bytes, u64 offset, int flags,
+			     int compress_type);
 void btrfs_add_ordered_sum(struct btrfs_ordered_extent *entry,
 			   struct btrfs_ordered_sum *sum);
 struct btrfs_ordered_extent *btrfs_lookup_ordered_extent(struct inode *inode,
-- 
2.25.1

[PATCH v4 6/9] btrfs: support different disk extent size for delalloc

[Omar Sandoval]

From: Omar Sandoval <osandov@fb.com>

Currently, we always reserve the same extent size in the file and extent
size on disk for delalloc because the former is the worst case for the
latter. For RWF_ENCODED writes, we know the exact size of the extent on
disk, which may be less than or greater than (for bookends) the size in
the file. Add a disk_num_bytes parameter to
btrfs_delalloc_reserve_metadata() so that we can reserve the correct
amount of csum bytes. Additionally, make
btrfs_free_reserve_data_space_noquota() take a number of bytes instead
of a range, as it refers to the extent size on disk, not in the file. No
functional change.

Reviewed-by: Nikolay Borisov <nborisov@suse.com>
Signed-off-by: Omar Sandoval <osandov@fb.com>
---
 fs/btrfs/ctree.h          |  3 ++-
 fs/btrfs/delalloc-space.c | 38 +++++++++++++++++---------------------
 fs/btrfs/delalloc-space.h |  4 ++--
 fs/btrfs/file.c           |  3 ++-
 fs/btrfs/inode.c          |  7 ++-----
 fs/btrfs/relocation.c     |  4 ++--
 6 files changed, 27 insertions(+), 32 deletions(-)

diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h
index d97983ab94fe..72d7c249179d 100644
--- a/fs/btrfs/ctree.h
+++ b/fs/btrfs/ctree.h
@@ -2540,7 +2540,8 @@ void btrfs_subvolume_release_metadata(struct btrfs_fs_info *fs_info,
 				      struct btrfs_block_rsv *rsv);
 void btrfs_delalloc_release_extents(struct btrfs_inode *inode, u64 num_bytes);
 
-int btrfs_delalloc_reserve_metadata(struct btrfs_inode *inode, u64 num_bytes);
+int btrfs_delalloc_reserve_metadata(struct btrfs_inode *inode, u64 num_bytes,
+				    u64 disk_num_bytes);
 u64 btrfs_account_ro_block_groups_free_space(struct btrfs_space_info *sinfo);
 int btrfs_error_unpin_extent_range(struct btrfs_fs_info *fs_info,
 				   u64 start, u64 end);
diff --git a/fs/btrfs/delalloc-space.c b/fs/btrfs/delalloc-space.c
index 4cdac4d834f5..7c4dfa76b8d6 100644
--- a/fs/btrfs/delalloc-space.c
+++ b/fs/btrfs/delalloc-space.c
@@ -153,34 +153,28 @@ int btrfs_check_data_free_space(struct inode *inode,
 	/* Use new btrfs_qgroup_reserve_data to reserve precious data space. */
 	ret = btrfs_qgroup_reserve_data(inode, reserved, start, len);
 	if (ret < 0)
-		btrfs_free_reserved_data_space_noquota(inode, start, len);
+		btrfs_free_reserved_data_space_noquota(fs_info, len);
 	else
 		ret = 0;
 	return ret;
 }
 
 /*
- * Called if we need to clear a data reservation for this inode
- * Normally in a error case.
+ * Called if we need to clear a data reservation, normally in an error case.
  *
  * This one will *NOT* use accurate qgroup reserved space API, just for case
  * which we can't sleep and is sure it won't affect qgroup reserved space.
  * Like clear_bit_hook().
  */
-void btrfs_free_reserved_data_space_noquota(struct inode *inode, u64 start,
-					    u64 len)
+void btrfs_free_reserved_data_space_noquota(struct btrfs_fs_info *fs_info,
+					    u64 num_bytes)
 {
-	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
 	struct btrfs_space_info *data_sinfo;
 
-	/* Make sure the range is aligned to sectorsize */
-	len = round_up(start + len, fs_info->sectorsize) -
-	      round_down(start, fs_info->sectorsize);
-	start = round_down(start, fs_info->sectorsize);
-
+	num_bytes = ALIGN(num_bytes, fs_info->sectorsize);
 	data_sinfo = fs_info->data_sinfo;
 	spin_lock(&data_sinfo->lock);
-	btrfs_space_info_update_bytes_may_use(fs_info, data_sinfo, -len);
+	btrfs_space_info_update_bytes_may_use(fs_info, data_sinfo, -num_bytes);
 	spin_unlock(&data_sinfo->lock);
 }
 
@@ -201,7 +195,7 @@ void btrfs_free_reserved_data_space(struct inode *inode,
 	      round_down(start, root->fs_info->sectorsize);
 	start = round_down(start, root->fs_info->sectorsize);
 
-	btrfs_free_reserved_data_space_noquota(inode, start, len);
+	btrfs_free_reserved_data_space_noquota(root->fs_info, len);
 	btrfs_qgroup_free_data(inode, reserved, start, len);
 }
 
@@ -280,11 +274,11 @@ static void btrfs_calculate_inode_block_rsv_size(struct btrfs_fs_info *fs_info,
 }
 
 static void calc_inode_reservations(struct btrfs_fs_info *fs_info,
-				    u64 num_bytes, u64 *meta_reserve,
-				    u64 *qgroup_reserve)
+				    u64 num_bytes, u64 disk_num_bytes,
+				    u64 *meta_reserve, u64 *qgroup_reserve)
 {
 	u64 nr_extents = count_max_extents(num_bytes);
-	u64 csum_leaves = btrfs_csum_bytes_to_leaves(fs_info, num_bytes);
+	u64 csum_leaves = btrfs_csum_bytes_to_leaves(fs_info, disk_num_bytes);
 	u64 inode_update = btrfs_calc_metadata_size(fs_info, 1);
 
 	*meta_reserve = btrfs_calc_insert_metadata_size(fs_info,
@@ -298,7 +292,8 @@ static void calc_inode_reservations(struct btrfs_fs_info *fs_info,
 	*qgroup_reserve = nr_extents * fs_info->nodesize;
 }
 
-int btrfs_delalloc_reserve_metadata(struct btrfs_inode *inode, u64 num_bytes)
+int btrfs_delalloc_reserve_metadata(struct btrfs_inode *inode, u64 num_bytes,
+				    u64 disk_num_bytes)
 {
 	struct btrfs_root *root = inode->root;
 	struct btrfs_fs_info *fs_info = root->fs_info;
@@ -328,6 +323,7 @@ int btrfs_delalloc_reserve_metadata(struct btrfs_inode *inode, u64 num_bytes)
 	}
 
 	num_bytes = ALIGN(num_bytes, fs_info->sectorsize);
+	disk_num_bytes = ALIGN(disk_num_bytes, fs_info->sectorsize);
 
 	/*
 	 * We always want to do it this way, every other way is wrong and ends
@@ -339,8 +335,8 @@ int btrfs_delalloc_reserve_metadata(struct btrfs_inode *inode, u64 num_bytes)
 	 * everything out and try again, which is bad.  This way we just
 	 * over-reserve slightly, and clean up the mess when we are done.
 	 */
-	calc_inode_reservations(fs_info, num_bytes, &meta_reserve,
-				&qgroup_reserve);
+	calc_inode_reservations(fs_info, num_bytes, disk_num_bytes,
+				&meta_reserve, &qgroup_reserve);
 	ret = btrfs_qgroup_reserve_meta_prealloc(root, qgroup_reserve, true);
 	if (ret)
 		return ret;
@@ -359,7 +355,7 @@ int btrfs_delalloc_reserve_metadata(struct btrfs_inode *inode, u64 num_bytes)
 	spin_lock(&inode->lock);
 	nr_extents = count_max_extents(num_bytes);
 	btrfs_mod_outstanding_extents(inode, nr_extents);
-	inode->csum_bytes += num_bytes;
+	inode->csum_bytes += disk_num_bytes;
 	btrfs_calculate_inode_block_rsv_size(fs_info, inode);
 	spin_unlock(&inode->lock);
 
@@ -463,7 +459,7 @@ int btrfs_delalloc_reserve_space(struct inode *inode,
 	ret = btrfs_check_data_free_space(inode, reserved, start, len);
 	if (ret < 0)
 		return ret;
-	ret = btrfs_delalloc_reserve_metadata(BTRFS_I(inode), len);
+	ret = btrfs_delalloc_reserve_metadata(BTRFS_I(inode), len, len);
 	if (ret < 0)
 		btrfs_free_reserved_data_space(inode, *reserved, start, len);
 	return ret;
diff --git a/fs/btrfs/delalloc-space.h b/fs/btrfs/delalloc-space.h
index 54466fbd7075..f847f0a80409 100644
--- a/fs/btrfs/delalloc-space.h
+++ b/fs/btrfs/delalloc-space.h
@@ -13,8 +13,8 @@ void btrfs_free_reserved_data_space(struct inode *inode,
 void btrfs_delalloc_release_space(struct inode *inode,
 				  struct extent_changeset *reserved,
 				  u64 start, u64 len, bool qgroup_free);
-void btrfs_free_reserved_data_space_noquota(struct inode *inode, u64 start,
-					    u64 len);
+void btrfs_free_reserved_data_space_noquota(struct btrfs_fs_info *fs_info,
+					    u64 num_bytes);
 void btrfs_delalloc_release_metadata(struct btrfs_inode *inode, u64 num_bytes,
 				     bool qgroup_free);
 int btrfs_delalloc_reserve_space(struct inode *inode,
diff --git a/fs/btrfs/file.c b/fs/btrfs/file.c
index a16da274c9aa..bf862f59b2a8 100644
--- a/fs/btrfs/file.c
+++ b/fs/btrfs/file.c
@@ -1667,7 +1667,8 @@ static noinline ssize_t btrfs_buffered_write(struct kiocb *iocb,
 
 		WARN_ON(reserve_bytes == 0);
 		ret = btrfs_delalloc_reserve_metadata(BTRFS_I(inode),
-				reserve_bytes);
+						      reserve_bytes,
+						      reserve_bytes);
 		if (ret) {
 			if (!only_release_metadata)
 				btrfs_free_reserved_data_space(inode,
diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
index 0d5b4e14f815..bcde9903d13b 100644
--- a/fs/btrfs/inode.c
+++ b/fs/btrfs/inode.c
@@ -2000,9 +2000,7 @@ void btrfs_clear_delalloc_extent(struct inode *vfs_inode,
 		if (root->root_key.objectid != BTRFS_DATA_RELOC_TREE_OBJECTID &&
 		    do_list && !(state->state & EXTENT_NORESERVE) &&
 		    (*bits & EXTENT_CLEAR_DATA_RESV))
-			btrfs_free_reserved_data_space_noquota(
-					&inode->vfs_inode,
-					state->start, len);
+			btrfs_free_reserved_data_space_noquota(fs_info, len);
 
 		percpu_counter_add_batch(&fs_info->delalloc_bytes, -len,
 					 fs_info->delalloc_batch);
@@ -7168,8 +7166,7 @@ static int btrfs_get_blocks_direct_write(struct extent_map **map,
 			 * use the existing or preallocated extent, so does not
 			 * need to adjust btrfs_space_info's bytes_may_use.
 			 */
-			btrfs_free_reserved_data_space_noquota(inode, start,
-							       len);
+			btrfs_free_reserved_data_space_noquota(fs_info, len);
 			goto skip_cow;
 		}
 	}
diff --git a/fs/btrfs/relocation.c b/fs/btrfs/relocation.c
index 995d4b8b1cfd..5316245a065e 100644
--- a/fs/btrfs/relocation.c
+++ b/fs/btrfs/relocation.c
@@ -3300,8 +3300,8 @@ static int relocate_file_extent_cluster(struct inode *inode,
 	index = (cluster->start - offset) >> PAGE_SHIFT;
 	last_index = (cluster->end - offset) >> PAGE_SHIFT;
 	while (index <= last_index) {
-		ret = btrfs_delalloc_reserve_metadata(BTRFS_I(inode),
-				PAGE_SIZE);
+		ret = btrfs_delalloc_reserve_metadata(BTRFS_I(inode), PAGE_SIZE,
+						      PAGE_SIZE);
 		if (ret)
 			goto out;
 
-- 
2.25.1

[PATCH v4 7/9] btrfs: optionally extend i_size in cow_file_range_inline()

[Omar Sandoval]

From: Omar Sandoval <osandov@fb.com>

Currently, an inline extent is always created after i_size is extended
from btrfs_dirty_pages(). However, for encoded writes, we only want to
update i_size after we successfully created the inline extent. Add an
update_i_size parameter to cow_file_range_inline() and
insert_inline_extent() and pass in the size of the extent rather than
determining it from i_size. Since the start parameter is always passed
as 0, get rid of it and simplify the logic in these two functions. While
we're here, let's document the requirements for creating an inline
extent.

Signed-off-by: Omar Sandoval <osandov@fb.com>
---
 fs/btrfs/inode.c | 94 +++++++++++++++++++++++-------------------------
 1 file changed, 44 insertions(+), 50 deletions(-)

diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
index bcde9903d13b..a45336214fbb 100644
--- a/fs/btrfs/inode.c
+++ b/fs/btrfs/inode.c
@@ -164,9 +164,10 @@ static int btrfs_init_inode_security(struct btrfs_trans_handle *trans,
 static int insert_inline_extent(struct btrfs_trans_handle *trans,
 				struct btrfs_path *path, int extent_inserted,
 				struct btrfs_root *root, struct inode *inode,
-				u64 start, size_t size, size_t compressed_size,
+				size_t size, size_t compressed_size,
 				int compress_type,
-				struct page **compressed_pages)
+				struct page **compressed_pages,
+				bool update_i_size)
 {
 	struct extent_buffer *leaf;
 	struct page *page = NULL;
@@ -175,7 +176,7 @@ static int insert_inline_extent(struct btrfs_trans_handle *trans,
 	struct btrfs_file_extent_item *ei;
 	int ret;
 	size_t cur_size = size;
-	unsigned long offset;
+	u64 i_size;
 
 	ASSERT((compressed_size > 0 && compressed_pages) ||
 	       (compressed_size == 0 && !compressed_pages));
@@ -190,7 +191,7 @@ static int insert_inline_extent(struct btrfs_trans_handle *trans,
 		size_t datasize;
 
 		key.objectid = btrfs_ino(BTRFS_I(inode));
-		key.offset = start;
+		key.offset = 0;
 		key.type = BTRFS_EXTENT_DATA_KEY;
 
 		datasize = btrfs_file_extent_calc_inline_size(cur_size);
@@ -229,12 +230,10 @@ static int insert_inline_extent(struct btrfs_trans_handle *trans,
 		btrfs_set_file_extent_compression(leaf, ei,
 						  compress_type);
 	} else {
-		page = find_get_page(inode->i_mapping,
-				     start >> PAGE_SHIFT);
+		page = find_get_page(inode->i_mapping, 0);
 		btrfs_set_file_extent_compression(leaf, ei, 0);
 		kaddr = kmap_atomic(page);
-		offset = offset_in_page(start);
-		write_extent_buffer(leaf, kaddr + offset, ptr, size);
+		write_extent_buffer(leaf, kaddr, ptr, size);
 		kunmap_atomic(kaddr);
 		put_page(page);
 	}
@@ -250,7 +249,12 @@ static int insert_inline_extent(struct btrfs_trans_handle *trans,
 	 * before we unlock the pages.  Otherwise we
 	 * could end up racing with unlink.
 	 */
-	BTRFS_I(inode)->disk_i_size = inode->i_size;
+	i_size = i_size_read(inode);
+	if (update_i_size && size > i_size) {
+		i_size_write(inode, size);
+		i_size = size;
+	}
+	BTRFS_I(inode)->disk_i_size = i_size;
 	ret = btrfs_update_inode(trans, root, inode);
 
 fail:
@@ -263,36 +267,31 @@ static int insert_inline_extent(struct btrfs_trans_handle *trans,
  * does the checks required to make sure the data is small enough
  * to fit as an inline extent.
  */
-static noinline int cow_file_range_inline(struct inode *inode, u64 start,
-					  u64 end, size_t compressed_size,
+static noinline int cow_file_range_inline(struct inode *inode, u64 size,
+					  size_t compressed_size,
 					  int compress_type,
-					  struct page **compressed_pages)
+					  struct page **compressed_pages,
+					  bool update_i_size)
 {
 	struct btrfs_root *root = BTRFS_I(inode)->root;
 	struct btrfs_fs_info *fs_info = root->fs_info;
 	struct btrfs_trans_handle *trans;
-	u64 isize = i_size_read(inode);
-	u64 actual_end = min(end + 1, isize);
-	u64 inline_len = actual_end - start;
-	u64 aligned_end = ALIGN(end, fs_info->sectorsize);
-	u64 data_len = inline_len;
+	u64 data_len = compressed_size ? compressed_size : size;
 	int ret;
 	struct btrfs_path *path;
 	int extent_inserted = 0;
 	u32 extent_item_size;
 
-	if (compressed_size)
-		data_len = compressed_size;
-
-	if (start > 0 ||
-	    actual_end > fs_info->sectorsize ||
+	/*
+	 * We can create an inline extent if it ends at or beyond the current
+	 * i_size, is no larger than a sector (decompressed), and the (possibly
+	 * compressed) data fits in a leaf and the configured maximum inline
+	 * size.
+	 */
+	if (size < i_size_read(inode) || size > fs_info->sectorsize ||
 	    data_len > BTRFS_MAX_INLINE_DATA_SIZE(fs_info) ||
-	    (!compressed_size &&
-	    (actual_end & (fs_info->sectorsize - 1)) == 0) ||
-	    end + 1 < isize ||
-	    data_len > fs_info->max_inline) {
+	    data_len > fs_info->max_inline)
 		return 1;
-	}
 
 	path = btrfs_alloc_path();
 	if (!path)
@@ -305,27 +304,18 @@ static noinline int cow_file_range_inline(struct inode *inode, u64 start,
 	}
 	trans->block_rsv = &BTRFS_I(inode)->block_rsv;
 
-	if (compressed_size && compressed_pages)
-		extent_item_size = btrfs_file_extent_calc_inline_size(
-		   compressed_size);
-	else
-		extent_item_size = btrfs_file_extent_calc_inline_size(
-		    inline_len);
-
-	ret = __btrfs_drop_extents(trans, root, inode, path,
-				   start, aligned_end, NULL,
-				   1, 1, extent_item_size, &extent_inserted);
+	extent_item_size = btrfs_file_extent_calc_inline_size(data_len);
+	ret = __btrfs_drop_extents(trans, root, inode, path, 0,
+				   fs_info->sectorsize, NULL, 1, 1,
+				   extent_item_size, &extent_inserted);
 	if (ret) {
 		btrfs_abort_transaction(trans, ret);
 		goto out;
 	}
 
-	if (isize > actual_end)
-		inline_len = min_t(u64, isize, actual_end);
-	ret = insert_inline_extent(trans, path, extent_inserted,
-				   root, inode, start,
-				   inline_len, compressed_size,
-				   compress_type, compressed_pages);
+	ret = insert_inline_extent(trans, path, extent_inserted, root, inode,
+				   size, compressed_size, compress_type,
+				   compressed_pages, update_i_size);
 	if (ret && ret != -ENOSPC) {
 		btrfs_abort_transaction(trans, ret);
 		goto out;
@@ -335,7 +325,7 @@ static noinline int cow_file_range_inline(struct inode *inode, u64 start,
 	}
 
 	set_bit(BTRFS_INODE_NEEDS_FULL_SYNC, &BTRFS_I(inode)->runtime_flags);
-	btrfs_drop_extent_cache(BTRFS_I(inode), start, aligned_end - 1, 0);
+	btrfs_drop_extent_cache(BTRFS_I(inode), 0, fs_info->sectorsize - 1, 0);
 out:
 	/*
 	 * Don't forget to free the reserved space, as for inlined extent
@@ -605,13 +595,15 @@ static noinline int compress_file_range(struct async_chunk *async_chunk)
 			/* we didn't compress the entire range, try
 			 * to make an uncompressed inline extent.
 			 */
-			ret = cow_file_range_inline(inode, start, end, 0,
-						    BTRFS_COMPRESS_NONE, NULL);
+			ret = cow_file_range_inline(inode, actual_end, 0,
+						    BTRFS_COMPRESS_NONE, NULL,
+						    false);
 		} else {
 			/* try making a compressed inline extent */
-			ret = cow_file_range_inline(inode, start, end,
+			ret = cow_file_range_inline(inode, actual_end,
 						    total_compressed,
-						    compress_type, pages);
+						    compress_type, pages,
+						    false);
 		}
 		if (ret <= 0) {
 			unsigned long clear_flags = EXTENT_DELALLOC |
@@ -994,9 +986,11 @@ static noinline int cow_file_range(struct inode *inode,
 	inode_should_defrag(BTRFS_I(inode), start, end, num_bytes, SZ_64K);
 
 	if (start == 0) {
+		u64 actual_end = min_t(u64, i_size_read(inode), end + 1);
+
 		/* lets try to make an inline extent */
-		ret = cow_file_range_inline(inode, start, end, 0,
-					    BTRFS_COMPRESS_NONE, NULL);
+		ret = cow_file_range_inline(inode, actual_end, 0,
+					    BTRFS_COMPRESS_NONE, NULL, false);
 		if (ret == 0) {
 			/*
 			 * We use DO_ACCOUNTING here because we need the
-- 
2.25.1

[PATCH v4 8/9] btrfs: implement RWF_ENCODED reads

[Omar Sandoval]

From: Omar Sandoval <osandov@fb.com>

There are 4 main cases:

1. Inline extents: we copy the data straight out of the extent buffer.
2. Hole/preallocated extents: we indicate the size of the extent
   starting from the read position; we don't need to copy zeroes.
3. Regular, uncompressed extents: we read the sectors we need directly
   from disk.
4. Regular, compressed extents: we read the entire compressed extent
   from disk and indicate what subset of the decompressed extent is in
   the file.

This initial implementation simplifies a few things that can be improved
in the future:

- We hold the inode lock during the operation.
- Cases 1, 3, and 4 allocate temporary memory to read into before
  copying out to userspace.
- Cases 3 and 4 do not implement repair yet.

Signed-off-by: Omar Sandoval <osandov@fb.com>
---
 fs/btrfs/ctree.h |   2 +
 fs/btrfs/file.c  |  12 +-
 fs/btrfs/inode.c | 456 +++++++++++++++++++++++++++++++++++++++++++++++
 3 files changed, 469 insertions(+), 1 deletion(-)

diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h
index 72d7c249179d..b6fede6c872b 100644
--- a/fs/btrfs/ctree.h
+++ b/fs/btrfs/ctree.h
@@ -2949,6 +2949,8 @@ int btrfs_run_delalloc_range(struct inode *inode, struct page *locked_page,
 int btrfs_writepage_cow_fixup(struct page *page, u64 start, u64 end);
 void btrfs_writepage_endio_finish_ordered(struct page *page, u64 start,
 					  u64 end, int uptodate);
+ssize_t btrfs_encoded_read(struct kiocb *iocb, struct iov_iter *iter);
+
 extern const struct dentry_operations btrfs_dentry_operations;
 
 /* ioctl.c */
diff --git a/fs/btrfs/file.c b/fs/btrfs/file.c
index bf862f59b2a8..d72d77e358e3 100644
--- a/fs/btrfs/file.c
+++ b/fs/btrfs/file.c
@@ -390,6 +390,16 @@ int btrfs_run_defrag_inodes(struct btrfs_fs_info *fs_info)
 	return 0;
 }
 
+static ssize_t btrfs_file_read_iter(struct kiocb *iocb, struct iov_iter *iter)
+{
+	if (iocb->ki_flags & IOCB_ENCODED) {
+		if (iocb->ki_flags & IOCB_NOWAIT)
+			return -EOPNOTSUPP;
+		return btrfs_encoded_read(iocb, iter);
+	}
+	return generic_file_read_iter(iocb, iter);
+}
+
 /* simple helper to fault in pages and copy.  This should go away
  * and be replaced with calls into generic code.
  */
@@ -3446,7 +3456,7 @@ static int btrfs_file_open(struct inode *inode, struct file *filp)
 
 const struct file_operations btrfs_file_operations = {
 	.llseek		= btrfs_file_llseek,
-	.read_iter      = generic_file_read_iter,
+	.read_iter      = btrfs_file_read_iter,
 	.splice_read	= generic_file_splice_read,
 	.write_iter	= btrfs_file_write_iter,
 	.mmap		= btrfs_file_mmap,
diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
index a45336214fbb..a7cd380479ff 100644
--- a/fs/btrfs/inode.c
+++ b/fs/btrfs/inode.c
@@ -10072,6 +10072,462 @@ void btrfs_set_range_writeback(struct extent_io_tree *tree, u64 start, u64 end)
 	}
 }
 
+static int encoded_iov_compression_from_btrfs(struct encoded_iov *encoded,
+					      unsigned int compress_type)
+{
+	switch (compress_type) {
+	case BTRFS_COMPRESS_NONE:
+		encoded->compression = ENCODED_IOV_COMPRESSION_NONE;
+		break;
+	case BTRFS_COMPRESS_ZLIB:
+		encoded->compression = ENCODED_IOV_COMPRESSION_ZLIB;
+		break;
+	case BTRFS_COMPRESS_LZO:
+		encoded->compression = ENCODED_IOV_COMPRESSION_LZO;
+		break;
+	case BTRFS_COMPRESS_ZSTD:
+		encoded->compression = ENCODED_IOV_COMPRESSION_ZSTD;
+		break;
+	default:
+		return -EIO;
+	}
+	return 0;
+}
+
+static ssize_t btrfs_encoded_read_inline(struct kiocb *iocb,
+					 struct iov_iter *iter, u64 start,
+					 u64 lockend,
+					 struct extent_state **cached_state,
+					 u64 extent_start, size_t count,
+					 struct encoded_iov *encoded,
+					 bool *unlocked)
+{
+	struct inode *inode = file_inode(iocb->ki_filp);
+	struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
+	struct btrfs_path *path;
+	struct extent_buffer *leaf;
+	struct btrfs_file_extent_item *item;
+	u64 ram_bytes;
+	unsigned long ptr;
+	void *tmp;
+	ssize_t ret;
+
+	path = btrfs_alloc_path();
+	if (!path) {
+		ret = -ENOMEM;
+		goto out;
+	}
+	ret = btrfs_lookup_file_extent(NULL, BTRFS_I(inode)->root, path,
+				       btrfs_ino(BTRFS_I(inode)), extent_start,
+				       0);
+	if (ret) {
+		if (ret > 0) {
+			/* The extent item disappeared? */
+			ret = -EIO;
+		}
+		goto out;
+	}
+	leaf = path->nodes[0];
+	item = btrfs_item_ptr(leaf, path->slots[0],
+			      struct btrfs_file_extent_item);
+
+	ram_bytes = btrfs_file_extent_ram_bytes(leaf, item);
+	ptr = btrfs_file_extent_inline_start(item);
+
+	encoded->len = (min_t(u64, extent_start + ram_bytes, inode->i_size) -
+			iocb->ki_pos);
+	ret = encoded_iov_compression_from_btrfs(encoded,
+				 btrfs_file_extent_compression(leaf, item));
+	if (ret)
+		goto out;
+	if (encoded->compression) {
+		size_t inline_size;
+
+		inline_size = btrfs_file_extent_inline_item_len(leaf,
+						btrfs_item_nr(path->slots[0]));
+		if (inline_size > count) {
+			ret = -ENOBUFS;
+			goto out;
+		}
+		count = inline_size;
+		encoded->unencoded_len = ram_bytes;
+		encoded->unencoded_offset = iocb->ki_pos - extent_start;
+	} else {
+		encoded->len = encoded->unencoded_len = count =
+			min_t(u64, count, encoded->len);
+		ptr += iocb->ki_pos - extent_start;
+	}
+
+	tmp = kmalloc(count, GFP_NOFS);
+	if (!tmp) {
+		ret = -ENOMEM;
+		goto out;
+	}
+	read_extent_buffer(leaf, tmp, ptr, count);
+	btrfs_free_path(path);
+	path = NULL;
+	unlock_extent_cached(io_tree, start, lockend, cached_state);
+	inode_unlock_shared(inode);
+	*unlocked = true;
+
+	ret = copy_encoded_iov_to_iter(encoded, iter);
+	if (ret)
+		goto out_free;
+	ret = copy_to_iter(tmp, count, iter);
+	if (ret != count)
+		ret = -EFAULT;
+out_free:
+	kfree(tmp);
+out:
+	btrfs_free_path(path);
+	return ret;
+}
+
+struct btrfs_encoded_read_private {
+	struct inode *inode;
+	wait_queue_head_t wait;
+	atomic_t pending;
+	bool uptodate;
+	bool skip_csum;
+};
+
+static bool btrfs_encoded_read_check_csums(struct btrfs_io_bio *io_bio)
+{
+	struct btrfs_encoded_read_private *priv = io_bio->bio.bi_private;
+	struct inode *inode = priv->inode;
+	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
+	u32 sectorsize = fs_info->sectorsize;
+	struct bio_vec *bvec;
+	struct bvec_iter_all iter_all;
+	u64 offset = 0;
+
+	if (priv->skip_csum)
+		return true;
+	bio_for_each_segment_all(bvec, &io_bio->bio, iter_all) {
+		unsigned int i, nr_sectors, pgoff;
+
+		nr_sectors = BTRFS_BYTES_TO_BLKS(fs_info, bvec->bv_len);
+		pgoff = bvec->bv_offset;
+		for (i = 0; i < nr_sectors; i++) {
+			int csum_pos;
+
+			csum_pos = BTRFS_BYTES_TO_BLKS(fs_info, offset);
+			if (__readpage_endio_check(inode, io_bio, csum_pos,
+						   bvec->bv_page, pgoff,
+						   io_bio->logical + offset,
+						   sectorsize))
+				return false;
+			offset += sectorsize;
+			pgoff += sectorsize;
+		}
+	}
+	return true;
+}
+
+static void btrfs_encoded_read_endio(struct bio *bio)
+{
+	struct btrfs_encoded_read_private *priv = bio->bi_private;
+	struct btrfs_io_bio *io_bio = btrfs_io_bio(bio);
+
+	if (bio->bi_status || !btrfs_encoded_read_check_csums(io_bio))
+		priv->uptodate = false;
+	if (!atomic_dec_return(&priv->pending))
+		wake_up(&priv->wait);
+	btrfs_io_bio_free_csum(io_bio);
+	bio_put(bio);
+}
+
+static bool btrfs_submit_encoded_read(struct bio *bio)
+{
+	struct btrfs_encoded_read_private *priv = bio->bi_private;
+	struct inode *inode = priv->inode;
+	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
+	blk_status_t status;
+
+	atomic_inc(&priv->pending);
+
+	if (!priv->skip_csum) {
+		status = btrfs_lookup_bio_sums(inode, bio,
+					       btrfs_io_bio(bio)->logical,
+					       NULL);
+		if (status)
+			goto out;
+	}
+
+	status = btrfs_bio_wq_end_io(fs_info, bio, BTRFS_WQ_ENDIO_DATA);
+	if (status)
+		goto out;
+
+	status = btrfs_map_bio(fs_info, bio, 0);
+out:
+	if (status) {
+		bio->bi_status = status;
+		bio_endio(bio);
+		return false;
+	}
+	return true;
+}
+
+static ssize_t btrfs_encoded_read_regular(struct kiocb *iocb,
+					  struct iov_iter *iter,
+					  u64 start, u64 lockend,
+					  struct extent_state **cached_state,
+					  u64 offset, u64 disk_io_size,
+					  size_t count,
+					  const struct encoded_iov *encoded,
+					  bool *unlocked)
+{
+	struct inode *inode = file_inode(iocb->ki_filp);
+	struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
+	struct btrfs_encoded_read_private priv = {
+		.inode = inode,
+		.wait = __WAIT_QUEUE_HEAD_INITIALIZER(priv.wait),
+		.pending = ATOMIC_INIT(1),
+		.uptodate = true,
+		.skip_csum = BTRFS_I(inode)->flags & BTRFS_INODE_NODATASUM,
+	};
+	struct page **pages;
+	unsigned long nr_pages, i;
+	struct bio *bio = NULL;
+	u64 cur;
+	size_t page_offset;
+	ssize_t ret;
+
+	nr_pages = DIV_ROUND_UP(disk_io_size, PAGE_SIZE);
+	pages = kcalloc(nr_pages, sizeof(struct page *), GFP_NOFS);
+	if (!pages)
+		return -ENOMEM;
+	for (i = 0; i < nr_pages; i++) {
+		pages[i] = alloc_page(GFP_NOFS | __GFP_HIGHMEM);
+		if (!pages[i]) {
+			ret = -ENOMEM;
+			goto out;
+		}
+	}
+
+	i = 0;
+	cur = 0;
+	while (cur < disk_io_size) {
+		size_t bytes = min_t(u64, disk_io_size - cur,
+				     PAGE_SIZE);
+
+		if (!bio) {
+			bio = btrfs_bio_alloc(offset + cur);
+			bio->bi_end_io = btrfs_encoded_read_endio;
+			bio->bi_private = &priv;
+			bio->bi_opf = REQ_OP_READ;
+			btrfs_io_bio(bio)->logical = start + cur;
+		}
+
+		if (bio_add_page(bio, pages[i], bytes, 0) < bytes) {
+			bool success;
+
+			success = btrfs_submit_encoded_read(bio);
+			bio = NULL;
+			if (!success)
+				break;
+			continue;
+		}
+		i++;
+		cur += bytes;
+	}
+
+	if (bio)
+		btrfs_submit_encoded_read(bio);
+	if (atomic_dec_return(&priv.pending))
+		io_wait_event(priv.wait, !atomic_read(&priv.pending));
+	if (!priv.uptodate) {
+		ret = -EIO;
+		goto out;
+	}
+
+	unlock_extent_cached(io_tree, start, lockend, cached_state);
+	inode_unlock_shared(inode);
+	*unlocked = true;
+
+	ret = copy_encoded_iov_to_iter(encoded, iter);
+	if (ret)
+		goto out;
+	if (encoded->compression) {
+		i = 0;
+		page_offset = 0;
+	} else {
+		i = (iocb->ki_pos - start) >> PAGE_SHIFT;
+		page_offset = (iocb->ki_pos - start) & (PAGE_SIZE - 1);
+	}
+	cur = 0;
+	while (cur < count) {
+		size_t bytes = min_t(size_t, count - cur,
+				     PAGE_SIZE - page_offset);
+
+		if (copy_page_to_iter(pages[i], page_offset, bytes,
+				      iter) != bytes) {
+			ret = -EFAULT;
+			goto out;
+		}
+		i++;
+		cur += bytes;
+		page_offset = 0;
+	}
+	ret = count;
+out:
+	for (i = 0; i < nr_pages; i++) {
+		if (pages[i])
+			put_page(pages[i]);
+	}
+	kfree(pages);
+	return ret;
+}
+
+ssize_t btrfs_encoded_read(struct kiocb *iocb, struct iov_iter *iter)
+{
+	struct inode *inode = file_inode(iocb->ki_filp);
+	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
+	struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
+	ssize_t ret;
+	size_t count;
+	u64 start, lockend, offset, disk_io_size;
+	struct extent_state *cached_state = NULL;
+	struct extent_map *em;
+	struct encoded_iov encoded = {};
+	bool unlocked = false;
+
+	ret = generic_encoded_read_checks(iocb, iter);
+	if (ret < 0)
+		return ret;
+	if (ret == 0)
+		return copy_encoded_iov_to_iter(&encoded, iter);
+	count = ret;
+
+	file_accessed(iocb->ki_filp);
+
+	inode_lock_shared(inode);
+
+	if (iocb->ki_pos >= inode->i_size) {
+		inode_unlock_shared(inode);
+		return copy_encoded_iov_to_iter(&encoded, iter);
+	}
+	start = ALIGN_DOWN(iocb->ki_pos, fs_info->sectorsize);
+	/*
+	 * We don't know how long the extent containing iocb->ki_pos is, but if
+	 * it's compressed we know that it won't be longer than this.
+	 */
+	lockend = start + BTRFS_MAX_UNCOMPRESSED - 1;
+
+	for (;;) {
+		struct btrfs_ordered_extent *ordered;
+
+		ret = btrfs_wait_ordered_range(inode, start,
+					       lockend - start + 1);
+		if (ret)
+			goto out_unlock_inode;
+		lock_extent_bits(io_tree, start, lockend, &cached_state);
+		ordered = btrfs_lookup_ordered_range(BTRFS_I(inode), start,
+						     lockend - start + 1);
+		if (!ordered)
+			break;
+		btrfs_put_ordered_extent(ordered);
+		unlock_extent_cached(io_tree, start, lockend, &cached_state);
+		cond_resched();
+	}
+
+	em = btrfs_get_extent(BTRFS_I(inode), NULL, 0, start,
+			      lockend - start + 1);
+	if (IS_ERR(em)) {
+		ret = PTR_ERR(em);
+		goto out_unlock_extent;
+	}
+
+	if (em->block_start == EXTENT_MAP_INLINE) {
+		u64 extent_start = em->start;
+
+		/*
+		 * For inline extents we get everything we need out of the
+		 * extent item.
+		 */
+		free_extent_map(em);
+		em = NULL;
+		ret = btrfs_encoded_read_inline(iocb, iter, start, lockend,
+						&cached_state, extent_start,
+						count, &encoded, &unlocked);
+		goto out;
+	}
+
+	/*
+	 * We only want to return up to EOF even if the extent extends beyond
+	 * that.
+	 */
+	encoded.len = (min_t(u64, extent_map_end(em), inode->i_size) -
+		       iocb->ki_pos);
+	if (em->block_start == EXTENT_MAP_HOLE ||
+	    test_bit(EXTENT_FLAG_PREALLOC, &em->flags)) {
+		offset = EXTENT_MAP_HOLE;
+		encoded.len = encoded.unencoded_len = count =
+			min_t(u64, count, encoded.len);
+	} else if (test_bit(EXTENT_FLAG_COMPRESSED, &em->flags)) {
+		offset = em->block_start;
+		/*
+		 * Bail if the buffer isn't large enough to return the whole
+		 * compressed extent.
+		 */
+		if (em->block_len > count) {
+			ret = -ENOBUFS;
+			goto out_em;
+		}
+		disk_io_size = count = em->block_len;
+		encoded.unencoded_len = em->ram_bytes;
+		encoded.unencoded_offset = iocb->ki_pos - em->orig_start;
+		ret = encoded_iov_compression_from_btrfs(&encoded,
+							 em->compress_type);
+		if (ret)
+			goto out_em;
+	} else {
+		offset = em->block_start + (start - em->start);
+		if (encoded.len > count)
+			encoded.len = count;
+		/*
+		 * Don't read beyond what we locked. This also limits the page
+		 * allocations that we'll do.
+		 */
+		disk_io_size = min(lockend + 1, iocb->ki_pos + encoded.len) - start;
+		encoded.len = encoded.unencoded_len = count =
+			start + disk_io_size - iocb->ki_pos;
+		disk_io_size = ALIGN(disk_io_size, fs_info->sectorsize);
+	}
+	free_extent_map(em);
+	em = NULL;
+
+	if (offset == EXTENT_MAP_HOLE) {
+		unlock_extent_cached(io_tree, start, lockend, &cached_state);
+		inode_unlock_shared(inode);
+		unlocked = true;
+		ret = copy_encoded_iov_to_iter(&encoded, iter);
+		if (ret)
+			goto out;
+		ret = iov_iter_zero(count, iter);
+		if (ret != count)
+			ret = -EFAULT;
+	} else {
+		ret = btrfs_encoded_read_regular(iocb, iter, start, lockend,
+						 &cached_state, offset,
+						 disk_io_size, count, &encoded,
+						 &unlocked);
+	}
+
+out:
+	if (ret >= 0)
+		iocb->ki_pos += encoded.len;
+out_em:
+	free_extent_map(em);
+out_unlock_extent:
+	if (!unlocked)
+		unlock_extent_cached(io_tree, start, lockend, &cached_state);
+out_unlock_inode:
+	if (!unlocked)
+		inode_unlock_shared(inode);
+	return ret;
+}
+
 #ifdef CONFIG_SWAP
 /*
  * Add an entry indicating a block group or device which is pinned by a
-- 
2.25.1

[PATCH v4 9/9] btrfs: implement RWF_ENCODED writes

[Omar Sandoval]

From: Omar Sandoval <osandov@fb.com>

The implementation resembles direct I/O: we have to flush any ordered
extents, invalidate the page cache, and do the io tree/delalloc/extent
map/ordered extent dance. From there, we can reuse the compression code
with a minor modification to distinguish the write from writeback. This
also creates inline extents when possible.

Now that read and write are implemented, this also sets the
FMODE_ENCODED_IO flag in btrfs_file_open().

Signed-off-by: Omar Sandoval <osandov@fb.com>
---
 fs/btrfs/compression.c  |   7 +-
 fs/btrfs/compression.h  |   6 +-
 fs/btrfs/ctree.h        |   2 +
 fs/btrfs/file.c         |  40 +++++--
 fs/btrfs/inode.c        | 243 +++++++++++++++++++++++++++++++++++++++-
 fs/btrfs/ordered-data.c |  12 +-
 fs/btrfs/ordered-data.h |   2 +
 7 files changed, 295 insertions(+), 17 deletions(-)

diff --git a/fs/btrfs/compression.c b/fs/btrfs/compression.c
index b66846272971..b9db1cb70d7e 100644
--- a/fs/btrfs/compression.c
+++ b/fs/btrfs/compression.c
@@ -377,7 +377,8 @@ static void end_compressed_bio_write(struct bio *bio)
 			bio->bi_status == BLK_STS_OK);
 	cb->compressed_pages[0]->mapping = NULL;
 
-	end_compressed_writeback(inode, cb);
+	if (cb->writeback)
+		end_compressed_writeback(inode, cb);
 	/* note, our inode could be gone now */
 
 	/*
@@ -413,7 +414,8 @@ blk_status_t btrfs_submit_compressed_write(struct inode *inode, u64 start,
 				 struct page **compressed_pages,
 				 unsigned long nr_pages,
 				 unsigned int write_flags,
-				 struct cgroup_subsys_state *blkcg_css)
+				 struct cgroup_subsys_state *blkcg_css,
+				 bool writeback)
 {
 	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
 	struct bio *bio = NULL;
@@ -437,6 +439,7 @@ blk_status_t btrfs_submit_compressed_write(struct inode *inode, u64 start,
 	cb->mirror_num = 0;
 	cb->compressed_pages = compressed_pages;
 	cb->compressed_len = compressed_len;
+	cb->writeback = writeback;
 	cb->orig_bio = NULL;
 	cb->nr_pages = nr_pages;
 
diff --git a/fs/btrfs/compression.h b/fs/btrfs/compression.h
index d253f7aa8ed5..b5a359c2c4b9 100644
--- a/fs/btrfs/compression.h
+++ b/fs/btrfs/compression.h
@@ -47,6 +47,9 @@ struct compressed_bio {
 	/* the compression algorithm for this bio */
 	int compress_type;
 
+	/* Whether this is a write for writeback. */
+	bool writeback;
+
 	/* number of compressed pages in the array */
 	unsigned long nr_pages;
 
@@ -94,7 +97,8 @@ blk_status_t btrfs_submit_compressed_write(struct inode *inode, u64 start,
 				  struct page **compressed_pages,
 				  unsigned long nr_pages,
 				  unsigned int write_flags,
-				  struct cgroup_subsys_state *blkcg_css);
+				  struct cgroup_subsys_state *blkcg_css,
+				  bool writeback);
 blk_status_t btrfs_submit_compressed_read(struct inode *inode, struct bio *bio,
 				 int mirror_num, unsigned long bio_flags);
 
diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h
index b6fede6c872b..956f4deaa544 100644
--- a/fs/btrfs/ctree.h
+++ b/fs/btrfs/ctree.h
@@ -2950,6 +2950,8 @@ int btrfs_writepage_cow_fixup(struct page *page, u64 start, u64 end);
 void btrfs_writepage_endio_finish_ordered(struct page *page, u64 start,
 					  u64 end, int uptodate);
 ssize_t btrfs_encoded_read(struct kiocb *iocb, struct iov_iter *iter);
+ssize_t btrfs_encoded_write(struct kiocb *iocb, struct iov_iter *from,
+			    struct encoded_iov *encoded);
 
 extern const struct dentry_operations btrfs_dentry_operations;
 
diff --git a/fs/btrfs/file.c b/fs/btrfs/file.c
index d72d77e358e3..2f8fbe43c1b4 100644
--- a/fs/btrfs/file.c
+++ b/fs/btrfs/file.c
@@ -1890,8 +1890,7 @@ static void update_time_for_write(struct inode *inode)
 		inode_inc_iversion(inode);
 }
 
-static ssize_t btrfs_file_write_iter(struct kiocb *iocb,
-				    struct iov_iter *from)
+static ssize_t btrfs_file_write_iter(struct kiocb *iocb, struct iov_iter *from)
 {
 	struct file *file = iocb->ki_filp;
 	struct inode *inode = file_inode(file);
@@ -1901,16 +1900,24 @@ static ssize_t btrfs_file_write_iter(struct kiocb *iocb,
 	u64 end_pos;
 	ssize_t num_written = 0;
 	const bool sync = iocb->ki_flags & IOCB_DSYNC;
+	struct encoded_iov encoded;
 	ssize_t err;
 	loff_t pos;
 	size_t count;
 	loff_t oldsize;
 	int clean_page = 0;
 
-	if (!(iocb->ki_flags & IOCB_DIRECT) &&
-	    (iocb->ki_flags & IOCB_NOWAIT))
+	if ((iocb->ki_flags & IOCB_NOWAIT) &&
+	    (!(iocb->ki_flags & IOCB_DIRECT) ||
+	     (iocb->ki_flags & IOCB_ENCODED)))
 		return -EOPNOTSUPP;
 
+	if (iocb->ki_flags & IOCB_ENCODED) {
+		err = copy_encoded_iov_from_iter(&encoded, from);
+		if (err)
+			return err;
+	}
+
 	if (iocb->ki_flags & IOCB_NOWAIT) {
 		if (!inode_trylock(inode))
 			return -EAGAIN;
@@ -1918,14 +1925,27 @@ static ssize_t btrfs_file_write_iter(struct kiocb *iocb,
 		inode_lock(inode);
 	}
 
-	err = generic_write_checks(iocb, from);
-	if (err <= 0) {
+	if (iocb->ki_flags & IOCB_ENCODED) {
+		err = generic_encoded_write_checks(iocb, &encoded);
+		if (err) {
+			inode_unlock(inode);
+			return err;
+		}
+		count = encoded.len;
+	} else {
+		err = generic_write_checks(iocb, from);
+		if (err < 0) {
+			inode_unlock(inode);
+			return err;
+		}
+		count = iov_iter_count(from);
+	}
+	if (count == 0) {
 		inode_unlock(inode);
 		return err;
 	}
 
 	pos = iocb->ki_pos;
-	count = iov_iter_count(from);
 	if (iocb->ki_flags & IOCB_NOWAIT) {
 		/*
 		 * We will allocate space in case nodatacow is not set,
@@ -1984,7 +2004,9 @@ static ssize_t btrfs_file_write_iter(struct kiocb *iocb,
 	if (sync)
 		atomic_inc(&BTRFS_I(inode)->sync_writers);
 
-	if (iocb->ki_flags & IOCB_DIRECT) {
+	if (iocb->ki_flags & IOCB_ENCODED) {
+		num_written = btrfs_encoded_write(iocb, from, &encoded);
+	} else if (iocb->ki_flags & IOCB_DIRECT) {
 		num_written = __btrfs_direct_write(iocb, from);
 	} else {
 		num_written = btrfs_buffered_write(iocb, from);
@@ -3450,7 +3472,7 @@ static loff_t btrfs_file_llseek(struct file *file, loff_t offset, int whence)
 
 static int btrfs_file_open(struct inode *inode, struct file *filp)
 {
-	filp->f_mode |= FMODE_NOWAIT;
+	filp->f_mode |= FMODE_NOWAIT | FMODE_ENCODED_IO;
 	return generic_file_open(inode, filp);
 }
 
diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
index a7cd380479ff..b4b954daf310 100644
--- a/fs/btrfs/inode.c
+++ b/fs/btrfs/inode.c
@@ -871,7 +871,7 @@ static noinline void submit_compressed_extents(struct async_chunk *async_chunk)
 				    ins.offset, async_extent->pages,
 				    async_extent->nr_pages,
 				    async_chunk->write_flags,
-				    async_chunk->blkcg_css)) {
+				    async_chunk->blkcg_css, true)) {
 			struct page *p = async_extent->pages[0];
 			const u64 start = async_extent->start;
 			const u64 end = start + async_extent->ram_size - 1;
@@ -2503,7 +2503,8 @@ static int btrfs_finish_ordered_io(struct btrfs_ordered_extent *ordered_extent)
 
 	if (!test_bit(BTRFS_ORDERED_NOCOW, &ordered_extent->flags) &&
 	    !test_bit(BTRFS_ORDERED_PREALLOC, &ordered_extent->flags) &&
-	    !test_bit(BTRFS_ORDERED_DIRECT, &ordered_extent->flags))
+	    !test_bit(BTRFS_ORDERED_DIRECT, &ordered_extent->flags) &&
+	    !test_bit(BTRFS_ORDERED_ENCODED, &ordered_extent->flags))
 		clear_new_delalloc_bytes = true;
 
 	freespace_inode = btrfs_is_free_space_inode(BTRFS_I(inode));
@@ -10528,6 +10529,244 @@ ssize_t btrfs_encoded_read(struct kiocb *iocb, struct iov_iter *iter)
 	return ret;
 }
 
+ssize_t btrfs_encoded_write(struct kiocb *iocb, struct iov_iter *from,
+			    struct encoded_iov *encoded)
+{
+	struct inode *inode = file_inode(iocb->ki_filp);
+	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
+	struct btrfs_root *root = BTRFS_I(inode)->root;
+	struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
+	struct extent_changeset *data_reserved = NULL;
+	struct extent_state *cached_state = NULL;
+	int compression;
+	size_t orig_count;
+	u64 start, end;
+	u64 num_bytes, ram_bytes, disk_num_bytes;
+	unsigned long nr_pages, i;
+	struct page **pages;
+	struct btrfs_key ins;
+	bool extent_reserved = false;
+	struct extent_map *em;
+	ssize_t ret;
+
+	switch (encoded->compression) {
+	case ENCODED_IOV_COMPRESSION_ZLIB:
+		compression = BTRFS_COMPRESS_ZLIB;
+		break;
+	case ENCODED_IOV_COMPRESSION_LZO:
+		compression = BTRFS_COMPRESS_LZO;
+		break;
+	case ENCODED_IOV_COMPRESSION_ZSTD:
+		compression = BTRFS_COMPRESS_ZSTD;
+		break;
+	default:
+		return -EINVAL;
+	}
+	if (encoded->encryption != ENCODED_IOV_ENCRYPTION_NONE)
+		return -EINVAL;
+
+	orig_count = iov_iter_count(from);
+
+	/* The extent size must be sane. */
+	if (encoded->unencoded_len > BTRFS_MAX_UNCOMPRESSED ||
+	    orig_count > BTRFS_MAX_COMPRESSED || orig_count == 0)
+		return -EINVAL;
+
+	/*
+	 * The compressed data must be smaller than the decompressed data.
+	 *
+	 * It's of course possible for data to compress to larger or the same
+	 * size, but the buffered I/O path falls back to no compression for such
+	 * data, and we don't want to break any assumptions by creating these
+	 * extents.
+	 *
+	 * Note that this is less strict than the current check we have that the
+	 * compressed data must be at least one sector smaller than the
+	 * decompressed data. We only want to enforce the weaker requirement
+	 * from old kernels that it is at least one byte smaller.
+	 */
+	if (orig_count >= encoded->unencoded_len)
+		return -EINVAL;
+
+	/* The extent must start on a sector boundary. */
+	start = iocb->ki_pos;
+	if (!IS_ALIGNED(start, fs_info->sectorsize))
+		return -EINVAL;
+
+	/*
+	 * The extent must end on a sector boundary. However, we allow a write
+	 * which ends at or extends i_size to have an unaligned length; we round
+	 * up the extent size and set i_size to the unaligned end.
+	 */
+	if (start + encoded->len < inode->i_size &&
+	    !IS_ALIGNED(start + encoded->len, fs_info->sectorsize))
+		return -EINVAL;
+
+	/* Finally, the offset in the unencoded data must be sector-aligned. */
+	if (!IS_ALIGNED(encoded->unencoded_offset, fs_info->sectorsize))
+		return -EINVAL;
+
+	num_bytes = ALIGN(encoded->len, fs_info->sectorsize);
+	ram_bytes = ALIGN(encoded->unencoded_len, fs_info->sectorsize);
+	end = start + num_bytes - 1;
+
+	/*
+	 * If the extent cannot be inline, the compressed data on disk must be
+	 * sector-aligned. For convenience, we extend it with zeroes if it
+	 * isn't.
+	 */
+	disk_num_bytes = ALIGN(orig_count, fs_info->sectorsize);
+	nr_pages = DIV_ROUND_UP(disk_num_bytes, PAGE_SIZE);
+	pages = kvcalloc(nr_pages, sizeof(struct page *), GFP_KERNEL_ACCOUNT);
+	if (!pages)
+		return -ENOMEM;
+	for (i = 0; i < nr_pages; i++) {
+		size_t bytes = min_t(size_t, PAGE_SIZE, iov_iter_count(from));
+		char *kaddr;
+
+		pages[i] = alloc_page(GFP_KERNEL_ACCOUNT | __GFP_HIGHMEM);
+		if (!pages[i]) {
+			ret = -ENOMEM;
+			goto out_pages;
+		}
+		kaddr = kmap(pages[i]);
+		if (copy_from_iter(kaddr, bytes, from) != bytes) {
+			kunmap(pages[i]);
+			ret = -EFAULT;
+			goto out_pages;
+		}
+		if (bytes < PAGE_SIZE)
+			memset(kaddr + bytes, 0, PAGE_SIZE - bytes);
+		kunmap(pages[i]);
+	}
+
+	for (;;) {
+		struct btrfs_ordered_extent *ordered;
+
+		ret = btrfs_wait_ordered_range(inode, start, num_bytes);
+		if (ret)
+			goto out_pages;
+		ret = invalidate_inode_pages2_range(inode->i_mapping,
+						    start >> PAGE_SHIFT,
+						    end >> PAGE_SHIFT);
+		if (ret)
+			goto out_pages;
+		lock_extent_bits(io_tree, start, end, &cached_state);
+		ordered = btrfs_lookup_ordered_range(BTRFS_I(inode), start,
+						     num_bytes);
+		if (!ordered &&
+		    !filemap_range_has_page(inode->i_mapping, start, end))
+			break;
+		if (ordered)
+			btrfs_put_ordered_extent(ordered);
+		unlock_extent_cached(io_tree, start, end, &cached_state);
+		cond_resched();
+	}
+
+	ret = btrfs_alloc_data_chunk_ondemand(BTRFS_I(inode), disk_num_bytes);
+	if (ret)
+		goto out_unlock;
+	ret = btrfs_qgroup_reserve_data(inode, &data_reserved, start,
+					num_bytes);
+	if (ret)
+		goto out_free_data_space;
+	ret = btrfs_delalloc_reserve_metadata(BTRFS_I(inode), num_bytes,
+					      disk_num_bytes);
+	if (ret)
+		goto out_qgroup_free_data;
+
+	/* Try an inline extent first. */
+	if (start == 0 && encoded->unencoded_len == encoded->len &&
+	    encoded->unencoded_offset == 0) {
+		ret = cow_file_range_inline(inode, encoded->len, orig_count,
+					    compression, pages, true);
+		if (ret <= 0) {
+			if (ret == 0)
+				ret = orig_count;
+			goto out_delalloc_release;
+		}
+	}
+
+	ret = btrfs_reserve_extent(root, disk_num_bytes, disk_num_bytes,
+				   disk_num_bytes, 0, 0, &ins, 1, 1);
+	if (ret)
+		goto out_delalloc_release;
+	extent_reserved = true;
+
+	em = create_io_em(inode, start, num_bytes,
+			  start - encoded->unencoded_offset, ins.objectid,
+			  ins.offset, ins.offset, ram_bytes, compression,
+			  BTRFS_ORDERED_COMPRESSED);
+	if (IS_ERR(em)) {
+		ret = PTR_ERR(em);
+		goto out_free_reserved;
+	}
+	free_extent_map(em);
+
+	ret = btrfs_add_ordered_extent(inode, start, num_bytes, ram_bytes,
+				       ins.objectid, ins.offset,
+				       encoded->unencoded_offset,
+				       (1 << BTRFS_ORDERED_ENCODED) |
+				       (1 << BTRFS_ORDERED_COMPRESSED),
+				       compression);
+	if (ret) {
+		btrfs_drop_extent_cache(BTRFS_I(inode), start, end, 0);
+		goto out_free_reserved;
+	}
+	btrfs_dec_block_group_reservations(fs_info, ins.objectid);
+
+	if (start + encoded->len > inode->i_size)
+		i_size_write(inode, start + encoded->len);
+
+	unlock_extent_cached(io_tree, start, end, &cached_state);
+
+	btrfs_delalloc_release_extents(BTRFS_I(inode), num_bytes);
+
+	if (btrfs_submit_compressed_write(inode, start, num_bytes, ins.objectid,
+					  ins.offset, pages, nr_pages, 0, NULL,
+					  false)) {
+		struct page *page = pages[0];
+
+		page->mapping = inode->i_mapping;
+		btrfs_writepage_endio_finish_ordered(page, start, end, 0);
+		page->mapping = NULL;
+		ret = -EIO;
+		goto out_pages;
+	}
+	ret = orig_count;
+	goto out;
+
+out_free_reserved:
+	btrfs_dec_block_group_reservations(fs_info, ins.objectid);
+	btrfs_free_reserved_extent(fs_info, ins.objectid, ins.offset, 1);
+out_delalloc_release:
+	btrfs_delalloc_release_extents(BTRFS_I(inode), num_bytes);
+	btrfs_delalloc_release_metadata(BTRFS_I(inode), disk_num_bytes,
+					ret < 0);
+out_qgroup_free_data:
+	if (ret < 0)
+		btrfs_qgroup_free_data(inode, data_reserved, start, num_bytes);
+out_free_data_space:
+	/*
+	 * If btrfs_reserve_extent() succeeded, then we already decremented
+	 * bytes_may_use.
+	 */
+	if (!extent_reserved)
+		btrfs_free_reserved_data_space_noquota(fs_info, disk_num_bytes);
+out_unlock:
+	unlock_extent_cached(io_tree, start, end, &cached_state);
+out_pages:
+	for (i = 0; i < nr_pages; i++) {
+		if (pages[i])
+			put_page(pages[i]);
+	}
+	kvfree(pages);
+out:
+	if (ret >= 0)
+		iocb->ki_pos += encoded->len;
+	return ret;
+}
+
 #ifdef CONFIG_SWAP
 /*
  * Add an entry indicating a block group or device which is pinned by a
diff --git a/fs/btrfs/ordered-data.c b/fs/btrfs/ordered-data.c
index 9a5f35d35fa9..e35a32a96467 100644
--- a/fs/btrfs/ordered-data.c
+++ b/fs/btrfs/ordered-data.c
@@ -445,9 +445,15 @@ void btrfs_remove_ordered_extent(struct inode *inode,
 	spin_lock(&btrfs_inode->lock);
 	btrfs_mod_outstanding_extents(btrfs_inode, -1);
 	spin_unlock(&btrfs_inode->lock);
-	if (root != fs_info->tree_root)
-		btrfs_delalloc_release_metadata(btrfs_inode, entry->num_bytes,
-						false);
+	if (root != fs_info->tree_root) {
+		u64 release;
+
+		if (test_bit(BTRFS_ORDERED_ENCODED, &entry->flags))
+			release = entry->disk_num_bytes;
+		else
+			release = entry->num_bytes;
+		btrfs_delalloc_release_metadata(btrfs_inode, release, false);
+	}
 
 	if (test_bit(BTRFS_ORDERED_DIRECT, &entry->flags))
 		percpu_counter_add_batch(&fs_info->dio_bytes, -entry->num_bytes,
diff --git a/fs/btrfs/ordered-data.h b/fs/btrfs/ordered-data.h
index ef528fef5841..ba7eec3fd152 100644
--- a/fs/btrfs/ordered-data.h
+++ b/fs/btrfs/ordered-data.h
@@ -61,6 +61,8 @@ enum {
 	BTRFS_ORDERED_TRUNCATED,
 	/* Regular IO for COW */
 	BTRFS_ORDERED_REGULAR,
+	/* RWF_ENCODED I/O */
+	BTRFS_ORDERED_ENCODED,
 };
 
 struct btrfs_ordered_extent {
-- 
2.25.1